A method and system for personalized recommendation of promotional items

Abstract

The application provides a personalized recommendation method and system for preferential commodities, and relates to the technical field of data processing.The method comprises the following steps: obtaining historical commodity browsing data of a target user; establishing a weighted undirected graph reflecting the interaction relationship between the target user and each browsing commodity; calculating the correlation score of the target user and each browsing commodity in the weighted undirected graph through a restart random walk algorithm with a dynamic restart probability; calculating the similarity between the commodity characteristics of each browsing commodity through the Euclidean distance, and determining the bridging score in an inverse proportional relationship with the similarity; determining the recommendation score of each browsing commodity according to the correlation score and the bridging score of the target user and each browsing commodity; calculating the commodity characteristic similarity between each preferential commodity and the browsing commodity with the highest recommendation score through the Mahalanobis distance; and pushing the preferential commodities to the target user in the order of the Mahalanobis distance from small to large. The recommended preferential commodities can meet the user preferences, and the diversity of the recommendation results is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for personalized recommendation of discounted products. Background Technology

[0002] Product recommendation is a technology that uses algorithms and user data to recommend products that users may be interested in. It helps users discover products of potential interest and improves their shopping experience by analyzing user browsing, clicks, and purchasing behavior, combined with product attributes and other user behaviors.

[0003] Product recommendations play a crucial role in connecting users and products amidst a sea of ​​information, solving the problem of choice difficulties for users in an information-overloaded environment. It not only improves user experience and helps users quickly find the products they need, but also increases platform sales by attracting users to purchase through personalized recommendations, thus optimizing the platform's revenue structure.

[0004] However, existing product recommendation methods are overly limited to users' historical purchase records, leading to excessive recommendations of products that are closely related to or closely related to past purchases. This results in overly concentrated and undiverse recommendations, which may cause users to lose interest in repetitive recommendations and thus reduce their stickiness to the platform. Summary of the Invention

[0005] To address the technical problem that existing product recommendation methods are overly limited to users' historical purchase records, leading to excessive recommendations of products that are closely related to or closely matched to past purchases, resulting in overly concentrated and undiverse recommendations, which may cause users to lose interest in repetitive recommendations and thus reduce their stickiness to the platform, this invention provides a personalized recommendation method and system for discounted products.

[0006] The technical solutions provided by the embodiments of the present invention are as follows:

[0007] First aspect

[0008] This invention provides a personalized recommendation method for discounted products, comprising:

[0009] S1: Obtain the target user's historical product browsing data, which includes multiple browsed products, product characteristics of each browsed product, browsing duration of each browsed product, and number of clicks on each browsed product;

[0010] S2: Based on historical product browsing data, establish a weighted undirected graph that reflects the interaction between the target user and each browsed product. The nodes of the weighted undirected graph include target user nodes and browsed product nodes. The edges between each node in the weighted undirected graph are the connections between the target user and each browsed product. The weight of each edge is the degree of preference obtained based on the number of clicks and browsing time of the target user on each browsed product.

[0011] S3: Based on the degree of preference, calculate the relevance score between the target user and each browsed product in the weighted undirected graph using a restart random walk algorithm with dynamic restart probability;

[0012] S4: Calculate the correlation between the product features of each browsed product using Euclidean distance, and determine the bridging score, which is inversely proportional to the correlation. The bridging score reflects the range of the target user's preference for different browsed products.

[0013] S5: Determine the recommendation score for each browsed product based on the relevance score and bridging score between the target user and each browsed product;

[0014] S6: Calculate the similarity of product features between each discounted product and the browsed product with the highest recommendation score using Mahalanobis distance;

[0015] S7: Push discounted products to target users in descending order of Mahal distance.

[0016] Second aspect

[0017] This invention provides a personalized recommendation system for discounted products, comprising:

[0018] processor;

[0019] A memory storing computer-readable instructions, which, when executed by the processor, implement the personalized recommendation method for discounted products as described in the first aspect.

[0020] Third aspect

[0021] The present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the personalized recommendation method for discounted products as described in the first aspect.

[0022] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0023] This invention acquires product browsing information from target users, not just those who have already purchased items, comprehensively capturing users' potential interests and actual needs. This improves the coverage and personalization of recommendations, satisfying diverse user preferences. A weighted undirected graph is built based on this data, combining click counts and browsing duration to capture user interests and preferences. Weights reflect the degree of preference for click counts and browsing durations, accurately capturing user interests and improving the personalization and accuracy of recommendations. Then, based on these preference levels, a restart random walk algorithm with dynamic restart probability is used to calculate the relevance score between the target user and each browsed product in the weighted undirected graph. Euclidean distance is then used to calculate the bridging score reflecting the range of the target user's preference for different browsed products. Assuming that each feature is independent, the correlation between product features of each browsed product is calculated, amplifying the differences in each feature individually. This helps to amplify the range of preference levels, emphasizing the diversity of user-browsed products during the recommendation process. Combining Euclidean distance to amplify the preference range accurately captures the differences in user interest in different products, emphasizing the diversity of recommended content, avoiding simplistic recommendations, and satisfying the diverse needs of users and their potential interests. Next, a recommendation score is determined for each browsed product based on relevance and bridging scores. Finally, Mahalanobis distance, which considers feature relevance, is used to calculate the similarity between the currently recommended discounted product and the browsed product with the highest recommendation score. Products are then recommended to the target user based on similarity. This effectively filters products that match user preferences and ensures the diversity and accuracy of recommendation results by emphasizing feature relevance. It effectively captures user preferences, ensuring recommendations match user preferences while improving the diversity of recommendation results, thereby increasing customer satisfaction and customer loyalty. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 A flowchart illustrating a personalized recommendation method for discounted products provided in an embodiment of the present invention;

[0026] Figure 2 This is a schematic diagram of the structure of a personalized recommendation system for discounted products provided in an embodiment of the present invention. Detailed Implementation

[0027] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0028] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0029] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0030] Reference manual attached Figure 1 The diagram shows a flowchart of a personalized recommendation method for discounted products provided by an embodiment of the present invention.

[0031] This invention provides a method for personalized recommendation of discounted products. This method can be implemented by a device for personalized recommendation of discounted products, which can be a terminal or a server. The processing flow of the method for personalized recommendation of discounted products may include the following steps:

[0032] S1: Obtain the target user's historical product browsing data.

[0033] Historical product browsing data includes multiple browsed products, product characteristics of each browsed product, browsing duration of each browsed product, and number of clicks on each browsed product.

[0034] It's important to note that by acquiring users' historical browsing data, including product characteristics, browsing duration, and click count, the system comprehensively records user interactions with products. By covering browsing behavior rather than just purchase records, it can capture users' potential interests and needs, expanding the scope of recommendations and improving the accuracy of personalized recommendations and user experience. Specifically, this target user is a single user—the individual who needs recommendations for discounted products.

[0035] S2: Build a weighted undirected graph based on historical product browsing data to reflect the interaction between the target user and each browsed product.

[0036] The nodes in the weighted undirected graph include target user nodes and browsed product nodes. The edges between the nodes in the weighted undirected graph are the connections between the target user and each browsed product. The weight of each edge is the degree of preference obtained based on the number of clicks and browsing time of the target user for each browsed product.

[0037] Weighted undirected graphs are graph structures where nodes represent objects (such as users and products), edges represent relationships between them, and edge weights quantify the strength or importance of these relationships. By constructing a weighted undirected graph using users' historical browsing data, target user nodes and browsed product nodes are connected by edges. The edge weights are calculated based on the number of clicks and browsing duration, reflecting the user's preference strength for each product. This process transforms complex user-product interactions into a mathematical graph model, providing structured foundational data for subsequent recommendation algorithms.

[0038] In one possible implementation, the weighted undirected graph is specifically:

[0039] G = (V, E, A)

[0040] A = (a ij )

[0041]

[0042] Where V represents the set of nodes including target user nodes and browsed product nodes, E represents the set of edges including the browsing time of each purchased product by the target user, A represents the target user's preference matrix for each browsed product, and a ij This indicates the degree of preference of target user i for browsing products, l ij c represents the browsing time of target user i for all browsed products, L represents the maximum browsing time of target user i for all browsed products. ij Let represent the number of clicks by target user i on the browsed products, C represent the maximum number of clicks by target user i on all browsed products, α1 and α2 represent the browsing duration weight and click count weight, respectively, H(L) and H(C) represent the entropy of browsing duration and click count, respectively, and φ 浏览商品 and φ 已购商品 Let V represent the set of browsed products and the set of purchased products, respectively. Let G represent a weighted undirected graph with respect to V, E, and W.

[0043] Specifically, the target user i can be any user who accesses the sales platform, while refers to the products that the user has browsed on the sales platform. Browsing products means that the user clicks on the product link corresponding to the product.

[0044] It's important to note that by connecting target users with the nodes and weights of browsed products, and comprehensively considering the weight allocation of browsing time and click count, a normalization method is used to accurately characterize the intensity of user preferences for different products. Simultaneously, by combining the edge weights and feature distribution information of users' historical interaction behavior, multi-dimensional features of user interests can be effectively captured. Furthermore, by dynamically balancing the influence weights of clicks and browsing, the flexibility and accuracy of preference modeling are ensured, providing a more structured and quantitative expression of preferences for subsequent recommendations. The advantage of preference intensity lies in its comprehensive integration of two behavioral features—browsing time and click count—and its normalization process, ensuring that the contribution of different feature dimensions is reasonably measured. Moreover, by dynamically allocating weights through information entropy when determining weights, the importance of features can be balanced, and adjustments can be made adaptively according to data characteristics, improving the accuracy and flexibility of preference expression.

[0045] Specifically, Among them, c i Let represent the number of clicks on the i-th viewed product, and let log represent the logarithmic function. c i The probability distribution, This represents the total number of clicks for all viewed products. Among them, l i This represents the browsing time for the i-th product. This represents the total browsing time for all viewed products. Optionally, α1 and α2 can each be directly set to 0.5.

[0046] S3: Based on preference levels, calculate the relevance scores between the target user and each browsed item in the weighted undirected graph using a restart random walk algorithm with dynamic restart probability.

[0047] The dynamic restart probability refers to the dynamic, rather than fixed, probability of returning to the starting point (target user node) at each step during a random walk. It adjusts based on the current walk state, allowing the system to flexibly balance local and global exploration. Dynamic restart probability can improve the diversity and accuracy of recommendations. The restart random walk algorithm is a method for calculating node relevance in a graph structure. Starting from the target node (user node), it has a certain probability of walking along edges to neighboring nodes at each step, and a certain probability of returning to the starting point. Through repeated walks, the algorithm can capture multi-level relationships between users and other nodes, outputting a relevance score. The relevance score represents the strength of the association between the target user node and each product node. It comprehensively considers the user's direct interest in the product (such as clicks and browsing time) and indirect relationships (such as interest in other similar products) to measure the product's attractiveness to the user.

[0048] By employing a dynamically restarted random walk algorithm, and utilizing a weighted undirected graph and dynamic restart probability, the relevance score between the target user and each product is calculated while balancing the exploration of local and latent interests. This process highlights products of current user interest while also uncovering diverse latent preferences, providing a foundation for accurate and varied recommendations.

[0049] In one possible implementation, the correlation score is calculated as follows:

[0050]

[0051] Among them, h q p represents the relevance score between the target user and the q-th viewed product. re e represents the restart probability. q Let be the unit vector describing the position of the q-th viewed product in the weighted undirected graph. Let T represent the normalized preference matrix, T0 represent the current number of steps, T0 represent the expected number of steps, β represent the adjustment coefficient, e represent the base of the natural logarithm, and D represent the degree matrix, where the diagonal elements of the degree matrix are the degrees of the nodes.

[0052] The degree of a node specifically refers to the sum of the weights of all edges associated with that node.

[0053] Where, unit vector e q It is a standard indicator vector whose function is to "select" the position of the q-th browsed product node in the weighted undirected graph, that is, to locate the position of the q-th node (browsed product) in the weighted undirected graph.

[0054] Here, the expected number of steps represents the system's expectation of completing local exploration within that number of steps. If T < T0 and the expected number of steps has not yet been reached, the system maintains a high restart probability; if T > T0 and the expected number of steps has been reached, the restart probability begins to decrease. Controlling the timing of switching between local and global exploration during random walks optimizes the personalized and diverse performance of the recommendation system. Optionally, the expected number of steps can be set to 1000, and the adjustment coefficient can be set to a value between 0.1 and 1, such as 0.5.

[0055] It's important to note that by combining dynamic restart probability and random walk calculations to determine relevance scores, and adjusting the restart probability based on the number of walk steps, the system balances local and global exploration. This ensures that the system initially prioritizes uncovering the user's current interests, then gradually expands its scope to explore potential interests, thereby improving the diversity and accuracy of recommendations. Using normalized preference and degree matrices effectively eliminates the impact of differences in the number of connections or weights between different nodes, allowing for a fairer quantification of the user-product interaction. This normalization ensures that the degree of preference reflects the true intensity of interest during calculation, while avoiding excessive bias from highly connected nodes. This optimizes the preference modeling of the recommendation system, improving the personalization effect and the fairness of the results.

[0056] S4: Calculate the similarity between the product features of each browsed product using Euclidean distance, and determine the bridging score, which is inversely proportional to the similarity.

[0057] The bridging score reflects the range of preferences of target users for different browsed products.

[0058] Euclidean distance is a mathematical tool used to measure the straight-line distance between two points (in this case, the product feature vectors). The bridging score is a score calculated inversely proportional to the similarity of product features, used to measure the span of a target user's preferences across different product categories. A larger span indicates user interest across categories, and the system will prioritize recommending more diverse products. A smaller span indicates concentrated user interests, and the bridging score helps avoid overly simplistic recommendations. By quantifying the differences in product features across browsing dimensions using Euclidean distance, and calculating the similarity between product features of each browsed product under the assumption that each feature is independent, the differences in each feature are amplified individually, helping to amplify the span of preference levels and emphasizing the diversity of products browsed by users during the recommendation process. The similarity between products is calculated, and the bridging score is derived based on the inverse proportional relationship, reflecting the span of user preferences across product categories. The bridging score improves diversity in recommendations, avoids overly concentrated recommendations, and strikes a balance between preference span and accuracy, providing users with recommendations that are both relevant to their interests and novel.

[0059] In one possible implementation, product characteristics include product price, product production date, and product type.

[0060] In one possible implementation, S4 specifically includes:

[0061] S401: Calculate the similarity between the product features of each browsed product:

[0062]

[0063] Among them, Ri [j, k] represents the similarity between the j-th and k-th browsed items viewed by the i-th target user, f j and f k Let denot j and k be the product feature vectors of the browsed products, respectively, and let σ represent the Gaussian kernel bandwidth parameter. exp represents the calculation of the square of the Euclidean distance, and exp represents the natural exponential function.

[0064] S402: Calculate the mean similarity:

[0065]

[0066] Among them, s i This represents the total number of products viewed by the i-th target user. This represents the total relevance score among the products viewed by the i-th target user. This represents the mean similarity between the products viewed by the i-th target user.

[0067] S403: Determine bridging score based on mean similarity:

[0068]

[0069] Among them, b i The bridging score represents the range of preferences of the i-th target user for different browsed products.

[0070] The bridging score reflects the characteristic span between the products browsed by the target user. A higher bridging score indicates a larger span between the products browsed by the target user, meaning weaker relevance. A higher bridging score makes it more likely to recommend products with different relevance to the user, resulting in more diverse recommendations. For example, identifying users connected to different product categories can lead to recommendations of multi-category products that they might be interested in.

[0071] It's important to note that by calculating the feature similarity between browsed products and combining it with a Gaussian kernel function, the feature span of user-preferred products can be quantified. A higher bridging score indicates greater diversity in the products browsed by the user, allowing the system to make recommendations for different product categories and improve the diversity of recommendation results. Simultaneously, the calculation of the mean similarity smooths out the influence of individual anomalous product features, ensuring the stability of the bridging score and providing an effective quantitative indicator for the recommendation system to find a balance between accuracy and diversity.

[0072] S5: Determine the recommendation score for each browsed product based on the relevance score and bridging score between the target user and each browsed product.

[0073] It should be noted that the recommendation score for each viewed product is calculated by combining the relevance score and bridging score between the target user and the viewed products. The relevance score reflects the user's current interests, while the bridging score measures the span of preferences. The combination of the two ensures that the recommendations are both accurate and diverse, balancing the user's personalized needs with the exploration of potential interests.

[0074] In one possible implementation, the recommendation score is calculated as follows:

[0075] t q =h q ×b i

[0076] Among them, t q h represents the recommendation score for the q-th viewed product. q b represents the relevance score between the target user and the q-th viewed product. i The bridging score represents the range of preferences of the i-th target user for different browsed products.

[0077] It's important to note that combining the relevance score and bridging score between users and products reflects both the user's direct preference for the product and the breadth of their browsing preferences, ensuring that recommendations are both accurate and diverse. In this way, the system can balance the mining of users' current and potential interests, enhancing the personalization and diversity of recommended content and meeting users' multi-dimensional needs.

[0078] S6: Calculate the similarity of product features between each discounted product and the browsed product with the highest recommendation score using Mahalanobis distance.

[0079] Mahalanobis distance is a distance metric for measuring the similarity between different features, taking into account the correlation between features. Mahalanobis distance normalizes features using the inverse of the covariance matrix, effectively handling the scale and correlation of different features. Using Mahalanobis distance, combined with the correlation and global distribution of product features, the similarity of candidate products can be accurately calculated. Product features often exhibit correlation (e.g., certain types of products have concentrated price ranges, or production dates and product categories are dependent). Mahalanobis distance captures the potential relationships between product features through the covariance matrix, making similarity calculations more accurate. The goal is to calculate the precise similarity between candidate discounted products and highly-rated products by users, providing high-quality matching for recommendation results.

[0080] In one possible implementation, the calculation method for product feature similarity is as follows:

[0081]

[0082] Among them, fv Let f represent the feature vector of the v-th discounted product. q1 This represents the feature vector of the browsed product with the highest recommendation score, where the subscript T indicates transpose. ∑ -1 d represents the inverse of the covariance matrix between the features of discounted products and the features of browsed products. M (f v f q1 ) represents f v with f q1 The Mahalanobis distance between the v-th discounted product and the browsed product with the highest recommendation score is the similarity of product features.

[0083] The feature dimensions of the discounted product feature vector are consistent with those of the product feature vector, namely, both include three feature dimensions: product price, product production date, and product type.

[0084] It should be noted that by utilizing Mahalanobis distance and combining it with the inverse of the covariance matrix, the correlation and distribution characteristics between different features are effectively considered, avoiding the problem of simple distance calculation ignoring the dependencies between features. By normalizing the features, the contribution of features at different scales to the similarity calculation is balanced, improving the accuracy of feature similarity calculation and thus ensuring a high degree of matching between the recommended product and the target product in the multi-dimensional feature space.

[0085] S7: Push discounted products to target users in ascending order of Mahal distance.

[0086] In one possible implementation, the process after S7 includes:

[0087] Historical product browsing data is updated at preset intervals.

[0088] Understandably, updating historical product browsing data at preset intervals can dynamically capture changes in user interests, ensuring that the recommendation system always makes personalized recommendations based on the latest user behavior, thereby improving the real-time performance and accuracy of the recommendation results.

[0089] It should be noted that those skilled in the art can set the preset duration according to actual needs, and this invention does not limit this.

[0090] In practical applications, the process begins by acquiring historical browsing data of the target user, including product features, click counts, and browsing duration, to comprehensively capture user interaction behavior. Next, a weighted undirected graph is constructed based on this data, quantifying the user's preference relationship with products using node and edge weights. Then, a dynamically restarted random walk algorithm is used to calculate the relevance score between the target user and each product, balancing the mining of local interests and potential preferences. Next, Euclidean distance is used to calculate the similarity of product features, generating a bridging score to broaden the range of user preferences and enhance recommendation diversity. Finally, the relevance score and bridging score are combined to determine the recommendation score for each browsed product. Then, Mahalanobis distance is used to calculate the similarity between discounted products and the product with the highest recommendation score, capturing potential associations between product features. Finally, discounted products are sorted by similarity from high to low and pushed to users, achieving accurate and diverse recommendation results.

[0091] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0092] This invention acquires product browsing information from target users, not just those who have already purchased items, comprehensively capturing users' potential interests and actual needs. This improves the coverage and personalization of recommendations, satisfying diverse user preferences. A weighted undirected graph is built based on this data, combining click counts and browsing duration to capture user interests and preferences. Weights reflect the degree of preference for click counts and browsing durations, accurately capturing user interests and improving the personalization and accuracy of recommendations. Then, based on the degree of preference, a restart random walk algorithm with dynamic restart probability is used to calculate the relevance score between the target user and each browsed product in the weighted undirected graph. Euclidean distance is then used to calculate the bridging score reflecting the range of the target user's preference for different browsed products. Assuming that each feature is independent, the similarity between the product features of each browsed product is calculated, amplifying the differences in each feature individually. This helps to amplify the range of preference levels and emphasize the diversity of user-browsed products during the recommendation process. Combining Euclidean distance to amplify the preference range accurately captures the differences in user interest in different products, emphasizing the diversity of recommended content, avoiding simplistic recommendations, and satisfying the diverse needs of users and their potential interests. Next, a recommendation score is determined for each browsed product based on relevance and bridging scores. Finally, Mahalanobis distance, which considers feature relevance, is used to calculate the similarity between the currently recommended discounted product and the browsed product with the highest recommendation score. Products are then recommended to the target user based on similarity. This effectively filters products that match user preferences and ensures the diversity and accuracy of recommendation results by emphasizing feature relevance. It effectively captures user preferences, ensuring recommendations match user preferences while improving the diversity of recommendation results, thereby increasing customer satisfaction and customer loyalty.

[0093] Reference manual attached Figure 2The diagram shows a structural schematic of a personalized recommendation system for discounted products provided by the present invention.

[0094] The present invention also provides a personalized recommendation system 20 for discounted products, applied to the above-mentioned personalized recommendation method for discounted products, comprising:

[0095] Processor 201.

[0096] The memory 202 stores computer-readable instructions, which, when executed by the processor 201, implement the personalized recommendation method for discounted products as described in the method embodiment.

[0097] The personalized recommendation system 20 for discounted products provided by this invention can execute the above-described personalized recommendation method for discounted products and achieve the same or similar technical effects. To avoid repetition, this invention will not elaborate further.

[0098] It should be understood that the processor in the embodiments of the present invention can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0099] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DRRAM).

[0100] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0101] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0102] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0103] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0104] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0105] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0106] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only 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 device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0107] 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 network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0108] In addition, the functional units in the various embodiments of the present invention 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.

[0109] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a 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 described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0110] This invention provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the personalized recommendation method for discounted products as described in the method embodiments.

[0111] The present invention provides a computer-readable storage medium that can implement the steps and effects of the personalized recommendation method for discounted products in the above-described method embodiments. To avoid repetition, the present invention will not repeat the details.

[0112] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0113] The following points need to be explained:

[0114] (1) The accompanying drawings of the embodiments of the present invention only involve the structures involved in the embodiments of the present invention. Other structures can refer to the general design.

[0115] (2) For clarity, the thickness of layers or regions is enlarged or reduced in the drawings used to describe embodiments of the present invention; that is, these drawings are not drawn to actual scale. It is understood that when an element such as a layer, film, region, or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element, or there may be intermediate elements.

[0116] (3) Where there is no conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0117] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for personalized recommendation of discounted products, characterized in that, include: S1: Obtain the target user's historical product browsing data, wherein the historical product browsing data includes multiple browsed products, product characteristics of each browsed product, browsing duration of each browsed product, and number of clicks on each browsed product; S2: Based on the historical product browsing data, establish a weighted undirected graph reflecting the interaction relationship between the target user and each browsed product. The nodes of the weighted undirected graph include target user nodes and browsed product nodes. The edges between each node in the weighted undirected graph are the connections between the target user and each browsed product. The weight of each edge is the degree of preference obtained based on the number of clicks and browsing time of the target user on each browsed product. S3: Based on the preference level, calculate the relevance score between the target user and each browsed product in the weighted undirected graph using a restart random walk algorithm with dynamic restart probability; S4: Calculate the similarity between the product features of each browsed product using Euclidean distance, and determine a bridging score that is inversely proportional to the similarity, wherein the bridging score reflects the range of the target user's preference for different browsed products; S5: Determine the recommendation score for each browsed product based on the relevance score between the target user and each browsed product and the bridging score; S6: Calculate the similarity of product features between each discounted product and the browsed product with the highest recommendation score using Mahalanobis distance; S7: Push the discounted products to the target users in ascending order of Mahalanobis distance; The specific method for calculating the correlation score is as follows: in, Indicates the target user and the first q The relevance score of each browsed product. Indicates the probability of restarting. Indicates the description of the first q The unit vector representing the position of a browsed product in a weighted undirected graph. This represents the normalized form of the preference matrix. T Indicates the current number of steps taken. T 0 represents the expected number of steps. β This represents the adjustment coefficient. e The base of the natural logarithm. D Let represent the degree matrix, where the diagonal elements of the degree matrix are the degrees of the nodes; Specifically, S4 includes: S401: Calculate the similarity between the product features of each browsed product: in, Indicates the first i The first target user browsed the j The first browsed product and the first k Similarity between browsed products and They represent the first j The and the first k The product feature vector of each browsed product. This represents the Gaussian kernel bandwidth parameter. exp represents the calculation of the square of the Euclidean distance; S402: Calculate the mean similarity: in, Indicates the first i The total number of products viewed by each target user. Indicates the first i The total relevance score between the products viewed by each target user. Indicates the first i The mean similarity between the products viewed by each target user; S403: Determine the bridging score based on the mean similarity: in, Indicates the first i The bridging score between the browsed items of the target user reflects the [missing information]. i The range of preferences of a target user for different browsed products; The specific method for calculating the similarity of product features is as follows: in, Indicates the first v Feature vectors of discounted products, This represents the feature vector of the browsed product with the highest recommendation score, indicated by the subscript. T Indicates transpose. This represents the inverse of the covariance matrix between the features of discounted products and the features of browsed products. express and The Mahalanobis distance between them is the first v The similarity of product features between each discounted product and the browsed product with the highest recommendation score; The specific method for calculating the recommendation score is as follows: in, Indicates the first q Recommendation score for each viewed product. Indicates the target user and the first q The relevance score of each browsed product. Indicates the first i The bridging score between the browsed items of the target user reflects the [missing information]. i The range of preferences of target users for different browsed products.

2. The personalized recommendation method for discounted products according to claim 1, characterized in that, The weighted undirected graph is specifically: in, V This represents a set of nodes that includes target user nodes and browsing product nodes. E This represents the set of edges that include the browsing time of each purchased item by the target user. A This represents a matrix showing the target user's preference for each browsed product. Indicates target user i For browsing products j The degree of preference, Indicates target user i For browsing products j Browsing duration L Indicates target user i Maximum browsing time for all viewed products. Indicates target user i For browsing products j Number of clicks C Indicates target user i The maximum number of clicks for all viewed products. and These represent the weighting of browsing time and the weighting of click count, respectively. and The entropy represents the browsing duration and the number of clicks, respectively. and These represent browsing the product collection and purchasing the product collection, respectively. G Indicates about V , E and W A weighted undirected graph.

3. The personalized recommendation method for discounted products according to claim 1, characterized in that, The product characteristics include product price, product production date, and product type.

4. The personalized recommendation method for discounted products according to claim 1, characterized in that, Following S7, the following is also included: The historical product browsing data is updated at preset intervals.

5. A personalized recommendation system for discounted products, characterized in that, include: processor; A memory storing computer-readable instructions, which, when executed by the processor, implement the personalized recommendation method for discounted products as described in any one of claims 1 to 4.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the personalized recommendation method for discounted products as described in any one of claims 1 to 4.

Images