A product recommendation method, device, equipment, storage medium and program product

Abstract

This invention discloses a product recommendation method, apparatus, device, storage medium, and program product. It searches for similar users among other users based on user similarity between the target user and other users, and determines an initial product recommendation list for the target user based on the product preference information of different similar users. This ensures that the recommended products are not limited to the target user's preferred products, and can effectively cover products of different types or new products when the product preference information of similar users involves different types of products. Furthermore, by further filtering the obtained initial product recommendation list based on the target user's behavioral information, the accuracy and diversity of the generated target product recommendation list can be effectively improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a product recommendation method, apparatus, electronic device, computer-readable storage medium, and computer program product. Background Technology

[0002] In recent years, with the continuous development of the internet and information age, users are faced with an ever-increasing number of products. How to quickly select the best product is a crucial issue for users, and personalized recommendation technology has become a key support for improving user experience. Existing product recommendation methods typically rely solely on the current relationship data between users and products. However, they struggle to recommend products of different types or new products, resulting in a lack of diversity in product recommendations. Summary of the Invention

[0003] This invention provides a product recommendation method, apparatus, device, storage medium, and program product to solve the technical problem that existing product recommendation methods are difficult to recommend for different types of products or new products, resulting in a lack of diversity in product recommendations.

[0004] To address the aforementioned technical problems, a first aspect of this invention provides a product recommendation method, comprising: Based on the user similarity between the target user and the other users, a similar user search is performed on the other users to obtain a preset number of similar users; wherein, the other users are the set of users other than the target user; Based on the user similarity and the product preference information of each similar user, an initial product recommendation list for the target user is obtained; Based on the preset product classification information, the target classification information of each initially recommended product in the initial product recommendation list is determined; Based on the target classification information and the target user's behavior information, the initial product recommendation list is filtered to generate a target product recommendation list and sent to the target user.

[0005] As a preferred embodiment, the step of searching for similar users among the remaining users based on the user similarity between the target user and other users to obtain a preset number of similar users specifically includes: Using the user similarity as the fitness value, a genetic algorithm is used to search for similar users among the remaining users to obtain a preset number of similar users; wherein, the user similarity is determined based on the similarity between the first feature information of the target user and the second feature information of the remaining users.

[0006] As a preferred embodiment, the step of using the user similarity as a fitness value and employing a genetic algorithm to search for similar users among the remaining users to obtain a preset number of similar users specifically includes: Using the user similarity as the fitness value, a roulette wheel selection algorithm is used to perform a selection operation on the current user population to obtain a first optimized user population; wherein, the initial user population is a population composed of the remaining users. Perform a crossover operation on the first optimized user population to obtain a second optimized user population; Based on the initial user population, a mutation operation is performed on the second optimized user population to obtain a third optimized user population; Using the third optimized user population as the current user population, repeat the aforementioned steps until the fitness value of the current user population converges; based on the fitness value of the current user population, obtain a preset number of similar users.

[0007] As a preferred embodiment, obtaining the initial product recommendation list for the target user based on the user similarity and the product preference information of each similar user specifically includes: Based on the product preference information, determine the first demand rating of each similar user for each product; Based on the user similarity, the first demand score, and the target user's average product demand score, a second demand score for each product is obtained from the target user. Based on the first demand score and the second demand score, target similar users related to the target user are determined from each of the similar users; Based on the second demand rating of the target similar users and the product association matrix of the target users, the initial product recommendation list is obtained; wherein, the product association matrix is ​​used to represent the actual demand rating of the target users for each product.

[0008] As a preferred embodiment, obtaining the initial product recommendation list based on the second demand score of the target similar users and the product association matrix of the target users specifically includes: Based on the sparsity of the product association matrix and a preset sparsity threshold, the user type of the target user is determined; wherein, the sparsity is determined based on the number of products in the product association matrix that do not have the actual demand rating. When the user type is a new user, the initial product recommendation list is determined based on the first initial recommended product whose second demand score of the target similar user is greater than the preset demand score threshold. When the user type is not a new user, the initial product recommendation list is determined based on the first initial recommended product and the products with clearly defined needs whose actual need score is greater than the need score threshold.

[0009] As a preferred embodiment, the step of filtering the initial product recommendation list based on the target classification information and the target user's behavioral information, generating a target product recommendation list, and sending it to the target user specifically includes: Based on the behavioral information, the number of times the target user exhibits product demand behavior for each of the target categorization information is obtained; The initial product recommendation list is filtered according to the preset target number of recommended products and the number of times the product demand behavior occurs, and the target product recommendation list is generated and sent to the target user.

[0010] A second aspect of the present invention provides a product recommendation device, comprising: The similar user acquisition module is used to perform a similar user search on the remaining users based on the user similarity between the target user and the remaining users, and obtain a preset number of similar users; wherein, the remaining users are a set of users other than the target user; The initial product recommendation list acquisition module is used to obtain the initial product recommendation list of the target user based on the product preference information of each of the similar users; The product classification module is used to determine the target classification information of each initially recommended product in the initial product recommendation list based on preset product classification information; The product recommendation module is used to filter the initial product recommendation list based on the target classification information and the target user's behavior information, generate a target product recommendation list, and send it to the target user.

[0011] A third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the product recommendation method described in any of the first aspects.

[0012] A fourth aspect of the present invention provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the product recommendation method described in any of the first aspects.

[0013] A fifth aspect of the present invention provides a computer program product, including a computer program / instructions, wherein when the computer program / instructions are executed by a processor, they implement the steps of the product recommendation method described in any of the first aspects.

[0014] Compared to existing technologies, the beneficial effects of this invention are that by searching for similar users among other users based on the user similarity between the target user and other users, and determining the initial product recommendation list for the target user based on the product preference information of different similar users, the recommended products are not limited to the target user's preferred products, and can be effectively covered when the product preference information of similar users involves different types of products or new products; in addition, by further filtering the obtained initial product recommendation list based on the target user's behavioral information, the accuracy and diversity of the generated target product recommendation list can be effectively improved. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the product recommendation method in an embodiment of the present invention; Figure 2 This is a schematic diagram of the product recommendation device in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the electronic device in an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 The first aspect of this invention provides a product recommendation method, comprising the following steps S1 to S4: Step S1: Based on the user similarity between the target user and the other users, perform a similar user search on the other users to obtain a preset number of similar users; wherein, the other users are a set of users other than the target user; Step S2: Based on the user similarity and the product preference information of each similar user, obtain the initial product recommendation list for the target user; Step S3: Based on the preset product classification information, determine the target classification information for each initially recommended product in the initial product recommendation list; Step S4: Based on the target classification information and the target user's behavior information, filter the initial product recommendation list, generate a target product recommendation list, and send it to the target user.

[0018] Specifically, to ensure that product recommendations align with the target user's preferences, this embodiment requires searching for similar users based on the user similarity between the target user and other users. It's understood that the users in this embodiment can be users of a product platform or shopping platform. Taking a cloud product platform as an example, the target user can be any user on that cloud product platform whose products are to be recommended, while the other users are all users on that cloud product platform other than the target user. To avoid limiting product recommendations to products favored by the target user, this embodiment considers not only the similarity of preferred products between the target user and other users, but also user similarity, thereby uncovering the target user's potential interests and recommending different types of products or new products. The number of similar users acquired in this embodiment can be set according to needs. For example, to ensure a wider coverage of product recommendations, the number of similar users acquired can be set to a larger number, such as 10, 15, or 18. This embodiment does not impose a specific limitation here.

[0019] Furthermore, this embodiment combines user-based collaborative filtering recommendation algorithms and content-based recommendation algorithms to generate the final target product recommendation list. First, this embodiment employs a user-based collaborative filtering recommendation algorithm. Based on the user similarity and product preference information of each similar user obtained above, it determines the preferred products of users with similar interests to the target user. It can be understood that for products that the target user has not yet viewed, users with similar interests may have viewed them or even have a need for them. Therefore, this step can uncover the target user's potential interests in products and form an initial product recommendation list. It is worth noting that the initial product recommendation list can include all preferred products of similar users related to the target user, or it can be filtered to include products that the target user has not yet used or purchased.

[0020] Furthermore, to improve the accuracy of product recommendations and ensure that new products with similar or identical interests to the target user are recommended to the target user in a timely manner, this embodiment pre-sets product classification information. This product classification information is formed by analyzing and calculating the characteristics of each product, and assigning the same classification label to products with similar characteristics. Taking cloud products as an example, the classification label can include cloud network products, cloud storage products, database products, and cloud security products, etc., while object storage products and cloud disk products can both be classified as cloud storage products. This embodiment will not elaborate further on this point. The classification of each product can be achieved by using a conventional classification model after model training. Based on this product classification information, the target classification information for each initially recommended product is determined.

[0021] Furthermore, based on the target user's behavioral information, it is possible to clarify which category of target information the target user's needs or interests lean towards. This information can then be used as the basis for filtering the initial product recommendation list. Taking cloud products as an example, if the target user's behavioral information clarifies that their needs or interests lean towards cloud storage products, then the recommendation ratio for products in the initial product recommendation list that also belong to cloud storage products but which the target user has not yet used or purchased will be increased to ensure the accuracy of the product recommendation results. If a new product also belongs to cloud storage products, it can be guaranteed that the new product will be recommended. To avoid other types of products not being recommended and hindering the target user from accessing other types of products, other products in the initial product recommendation list that do not belong to cloud storage products and which the target user has not yet used or purchased will be recommended with a smaller recommendation ratio to ensure the diversity of the product recommendation results.

[0022] The product recommendation method provided in this invention searches for similar users among other users based on the user similarity between the target user and other users, and determines the initial product recommendation list for the target user based on the product preference information of different similar users. This ensures that the recommended products are not limited to the target user's preferred products, and can effectively cover products of different types or new products when the product preference information of similar users involves different types of products or new products. In addition, the initial product recommendation list is further filtered based on the target user's behavioral information, thereby effectively improving the accuracy and diversity of the generated target product recommendation list.

[0023] As a preferred embodiment, the step of searching for similar users among the remaining users based on the user similarity between the target user and other users to obtain a preset number of similar users specifically includes: Using the user similarity as the fitness value, a genetic algorithm is used to search for similar users among the remaining users to obtain a preset number of similar users; wherein, the user similarity is determined based on the similarity between the first feature information of the target user and the second feature information of the remaining users.

[0024] Specifically, since there may be relatively few products that users have a common need for, especially in scenarios with a large number of products, the product recommendation results generated by using only user-based collaborative filtering recommendation algorithms and content-based recommendation algorithms may have certain errors. Therefore, this embodiment uses a genetic algorithm to search for similar users. The user similarity is determined based on the similarity between the target user's first feature information and the second feature information of the other users, as shown in the following expression: ; in, This represents the i-th feature factor in the first feature information of target user a. This represents the i-th feature factor in the second feature information of user b among the remaining users, where I represents the total number of feature factors. This represents the similarity between the first feature information of target user a and the second feature information of user b.

[0025] It is worth noting that the user's characteristic factors in this embodiment may include the reason for migrating to the cloud, problems encountered during the migration process, business continuity requirements, user costs, cloud payment methods, the proportion of business resources migrated to the cloud, technical capability information, and business characteristic information. For example, the reasons for migrating to the cloud may include high infrastructure operation and maintenance difficulty, high resource expansion difficulty, high facility costs, and low security compliance; business continuity requirements include basic SLA (Service Level Agreement) requirements, the business's tolerance for interruption, and whether there is a need for cross-regional disaster recovery, etc.; cloud payment methods may include pay-as-you-go or one-time payment methods; technical capability information includes: whether the user adopts a fully managed cloud approach or a hybrid cloud approach, whether DevOps-related capabilities are required, and whether there are compatibility requirements for specific technology stacks, etc.; business characteristic information includes: stable business traffic or specific peak fluctuations, business growth in a specific future time period, and whether there is a need for multi-regional deployment. It is understood that different quantitative values ​​can be provided for users to choose from for the above description. For example, for cloud payment methods, the value "1" represents pay-as-you-go payment and the value "2" represents one-time payment. For reasons for cloud migration, different values ​​can correspond to different reasons. This embodiment does not make specific limitations here, so as to facilitate the transformation of the above information into feature factors and facilitate the calculation of similarity of user feature information.

[0026] The user similarity calculated above is used as the fitness value for each of the remaining users. The higher the user similarity between the target user and the other users, the higher the fitness value of the corresponding user. Based on the selection, crossover, and mutation operations of the genetic algorithm, the fitness values ​​are further used to select and eliminate individual users from the remaining users. Among them, the selection operation can improve the accuracy of product recommendations, while the crossover and mutation operations can improve the diversity of product recommendations and avoid the "Matthew effect" to some extent.

[0027] As a preferred embodiment, the step of using the user similarity as a fitness value and employing a genetic algorithm to search for similar users among the remaining users to obtain a preset number of similar users specifically includes: Using the user similarity as the fitness value, a roulette wheel selection algorithm is used to perform a selection operation on the current user population to obtain a first optimized user population; wherein, the initial user population is a population composed of the remaining users. Perform a crossover operation on the first optimized user population to obtain a second optimized user population; Based on the initial user population, a mutation operation is performed on the second optimized user population to obtain a third optimized user population; Using the third optimized user population as the current user population, repeat the aforementioned steps until the fitness value of the current user population converges; based on the fitness value of the current user population, obtain a preset number of similar users.

[0028] Specifically, this embodiment employs the roulette wheel selection algorithm during the selection process. The core idea of ​​the roulette wheel algorithm is that the probability of a user being selected is proportional to their fitness value, thus preventing users with low fitness values ​​from being directly eliminated. The roulette wheel algorithm involves individual selection probability, cumulative probability, and individual selection strategy. The calculation of individual selection probability is as follows: ; The cumulative probability is calculated as follows: ; in, This represents the individual selection probability of the i-th user in the current user population; represents the fitness value of the i-th user individual; m represents the number of users in the current user population; This represents the cumulative probability from the first user to the ith user.

[0029] Furthermore, the individual selection strategy involves randomly selecting a number from the interval [0,1] and observing which cumulative probability interval the random number falls into. For users with higher fitness values, the range covered within the interval is larger, so the probability of the random number falling into its corresponding cumulative probability interval is higher. Conversely, for users with lower fitness values, the range covered within the interval is smaller, so the probability of the random number falling into its corresponding cumulative probability interval is lower, but there is still a possibility of selection. This approach avoids, to some extent, the problem of users with lower fitness values ​​being directly eliminated.

[0030] Furthermore, the crossover operation involves exchanging the genetic information of two individuals. In this embodiment, the genetic information is the characteristic information of the user individuals. This embodiment uses a single-point crossover method, that is, two user individuals are randomly selected at a crossover point to be divided into characteristic information A1, characteristic information A2 and characteristic information B1, characteristic information B2. Then, the two newly generated user individuals are characteristic information A1 + characteristic information B2 and characteristic information B1 + characteristic information A2, respectively. The specific expression of the crossover operation is: ; in, This refers to the first new user individual; This indicates a second new user individual; and These represent the two parts of feature information segmented from the first user individual; and These represent the two parts of the second user individual; This represents the cross coefficient.

[0031] Furthermore, in this embodiment, a mutation operation is performed on the second optimized user population obtained through crossover to prevent the genetic algorithm from getting stuck in local optima and achieve a better global optimum. Specifically, the mutation operation is to replace the user individuals in the second optimized user population with a random user individual from the initial user population within the solution range, thereby obtaining a third optimized user population to maintain the diversity of the entire user population.

[0032] Using the third optimized user population as the current user population, the aforementioned genetic algorithm steps are re-executed until the fitness values ​​of the user population converge. For example, whether the total fitness value of each individual user in the user population converges can be used as the criterion for stopping the execution of the genetic algorithm, or whether the average fitness value of each individual user in the user population converges can be used as the criterion for stopping the execution of the genetic algorithm. This embodiment does not make specific limitations here.

[0033] As a preferred embodiment, obtaining the initial product recommendation list for the target user based on the user similarity and the product preference information of each similar user specifically includes: Based on the product preference information, determine the first demand rating of each similar user for each product; Based on the user similarity, the first demand score, and the target user's average product demand score, a second demand score for each product is obtained from the target user. Based on the first demand score and the second demand score, target similar users related to the target user are determined from each of the similar users; Based on the second demand rating of the target similar users and the product association matrix of the target users, the initial product recommendation list is obtained; wherein, the product association matrix is ​​used to represent the actual demand rating of the target users for each product.

[0034] Specifically, the product preference information records the first demand ratings of similar users for each product. The higher the rating, the greater the similar user's demand or interest in that product. Then, the potential interests of the target user are mined using the following expression to obtain the target user's second demand ratings for each product: ; in, This represents the second requirement rating of target user a for product r; N represents the total number of similar users; This represents the user similarity between target user a and the nth similar user; This represents the first demand rating of the nth similar user for product r; This represents the average product demand rating of target user a, which can be obtained by averaging the actual demand ratings of target user a for each product.

[0035] Furthermore, this embodiment uses the Pearson correlation coefficient to evaluate the correlation between the first demand score and the second demand score, thereby identifying the most relevant target similar users, as shown in the following expression: ; ; ; ; ; ; Where X represents the second demand rating vector of target user a. For example, if target user a's second demand ratings for product A, product B and product C are 5 points, 4 points and 3 points respectively, then X=(5,4,3); Y represents the first demand rating vector of similar users. This represents the standard deviation of the second demand rating vector; This represents the standard deviation of the first demand rating vector; This represents the average value of the second demand rating vector; denoted as the average value of the first demand rating vector; m is the total number of demand ratings.

[0036] It is worth noting that target similar users are those most relevant to the target user among all similar users. Their preferred products can, to some extent, reflect the target user's interests. Therefore, considering the various products they currently need and the target user's actual needs rating for each product, an initial product recommendation list is determined.

[0037] As a preferred embodiment, obtaining the initial product recommendation list based on the second demand score of the target similar users and the product association matrix of the target users specifically includes: Based on the sparsity of the product association matrix and a preset sparsity threshold, the user type of the target user is determined; wherein, the sparsity is determined based on the number of products in the product association matrix that do not have the actual demand rating. When the user type is a new user, the initial product recommendation list is determined based on the first initial recommended product whose second demand score of the target similar user is greater than the preset demand score threshold. When the user type is not a new user, the initial product recommendation list is determined based on the first initial recommended product and the products with clearly defined needs whose actual need score is greater than the need score threshold.

[0038] Specifically, this embodiment determines whether a target user is a new user based on the sparsity of the product association matrix. It can be understood that if the product association matrix has a larger number of products without actual demand ratings, it indicates that the target user has little contact with the products on the current platform and can be identified as a new user; conversely, it can be identified as an old user. Considering that the number of products will continue to increase, if the judgment is based on a preset threshold for the number of products without actual demand ratings, the threshold needs to be updated frequently. Therefore, the ratio of the number of products without actual demand ratings to the current total number of products can be used as the sparsity. Accordingly, a threshold for the ratio of the number of products without actual demand ratings to the total number of products is set as the sparsity threshold, thereby avoiding the need to update the sparsity threshold frequently.

[0039] Furthermore, when the user type is a new user, since they have less exposure to products, it is necessary to recommend products that they may be interested in to the target user as much as possible. The first initial recommended product is the first product recommended product whose second demand score is greater than the preset demand score threshold for similar users, thus determining the initial product recommendation list.

[0040] When a user is not a new user, it means that they have been exposed to many products and have a clear need for products with actual need scores greater than the need score threshold. It is not very meaningful to recommend these products to them again. Therefore, based on the products that do not belong to the products with clear needs among the first initial recommended products for users with similar targets whose second need scores are greater than the preset need score threshold, the initial product recommendation list is determined.

[0041] As a preferred embodiment, the step of filtering the initial product recommendation list based on the target classification information and the target user's behavioral information, generating a target product recommendation list, and sending it to the target user specifically includes: Based on the behavioral information, the number of times the target user exhibits product demand behavior for each of the target categorization information is obtained; The initial product recommendation list is filtered according to the preset target number of recommended products and the number of times the product demand behavior occurs, and the target product recommendation list is generated and sent to the target user.

[0042] Specifically, in order to determine which product category the target user is more interested in and to ensure the accuracy of the product recommendation results, this embodiment needs to obtain the number of product demand behaviors of the target user for each target category information. Among them, product demand behaviors include product browsing behavior, product attention behavior, product purchase behavior, etc., which will not be elaborated on in this embodiment.

[0043] Furthermore, to avoid recommending too many products to the target user and affecting their filtering experience, this embodiment pre-sets a target number of recommended products. If the number of products in the current initial product recommendation list exceeds this target number, further filtering is required based on the number of product demand behaviors of the target user. A threshold for the number of product demand behaviors can be pre-set. Target categorization information with a product demand behavior frequency greater than this threshold is considered preference categorization information. Therefore, the proportion of products belonging to this preference categorization information in the initial product recommendation list can be increased, while products not belonging to this preference categorization information are recommended at a lower proportion. Ultimately, the number of products in the target product recommendation list is equal to the target number of recommended products, thus ensuring both the accuracy and diversity of product recommendations.

[0044] Please see Figure 2 A second aspect of the present invention provides a product recommendation device 100, comprising: The similar user acquisition module 11 is used to perform a similar user search on the remaining users based on the user similarity between the target user and the remaining users, and obtain a preset number of similar users; wherein, the remaining users are a set of users other than the target user; The initial product recommendation list acquisition module 12 is used to obtain the initial product recommendation list of the target user based on the product preference information of each of the similar users; Product classification module 13 is used to determine the target classification information of each initially recommended product in the initial product recommendation list based on preset product classification information; Product recommendation module 14 is used to filter the initial product recommendation list based on the target classification information and the target user's behavior information, generate a target product recommendation list, and send it to the target user.

[0045] As a preferred embodiment, the similar user acquisition module 11 is used to perform a similar user search on the remaining users based on the user similarity between the target user and the remaining users, to obtain a preset number of similar users, specifically including: Using the user similarity as the fitness value, a genetic algorithm is used to search for similar users among the remaining users to obtain a preset number of similar users; wherein, the user similarity is determined based on the similarity between the first feature information of the target user and the second feature information of the remaining users.

[0046] As a preferred embodiment, the similar user acquisition module 11 is used to perform a similar user search on the remaining users using the user similarity as the fitness value and a genetic algorithm to obtain a preset number of similar users, specifically including: Using the user similarity as the fitness value, a roulette wheel selection algorithm is used to perform a selection operation on the current user population to obtain a first optimized user population; wherein, the initial user population is a population composed of the remaining users. Perform a crossover operation on the first optimized user population to obtain a second optimized user population; Based on the initial user population, a mutation operation is performed on the second optimized user population to obtain a third optimized user population; Using the third optimized user population as the current user population, repeat the aforementioned steps until the fitness value of the current user population converges; based on the fitness value of the current user population, obtain a preset number of similar users.

[0047] As a preferred embodiment, the initial product recommendation list acquisition module 12 is used to obtain the initial product recommendation list of the target user based on the user similarity and the product preference information of each of the similar users, specifically including: Based on the product preference information, determine the first demand rating of each similar user for each product; Based on the user similarity, the first demand score, and the target user's average product demand score, a second demand score for each product is obtained from the target user. Based on the first demand score and the second demand score, target similar users related to the target user are determined from each of the similar users; Based on the second demand rating of the target similar users and the product association matrix of the target users, the initial product recommendation list is obtained; wherein, the product association matrix is ​​used to represent the actual demand rating of the target users for each product.

[0048] As a preferred embodiment, the initial product recommendation list acquisition module 12 is used to obtain the initial product recommendation list based on the second demand score of the target similar users and the product association matrix of the target users, specifically including: Based on the sparsity of the product association matrix and a preset sparsity threshold, the user type of the target user is determined; wherein, the sparsity is determined based on the number of products in the product association matrix that do not have the actual demand rating. When the user type is a new user, the initial product recommendation list is determined based on the first initial recommended product whose second demand score of the target similar user is greater than the preset demand score threshold. When the user type is not a new user, the initial product recommendation list is determined based on the first initial recommended product and the products with clearly defined needs whose actual need score is greater than the need score threshold.

[0049] As a preferred embodiment, the product recommendation module 14 is used to filter the initial product recommendation list based on the target classification information and the target user's behavior information, generate a target product recommendation list, and send it to the target user, specifically including: Based on the behavioral information, the number of times the target user exhibits product demand behavior for each of the target categorization information is obtained; The initial product recommendation list is filtered according to the preset target number of recommended products and the number of times the product demand behavior occurs, and the target product recommendation list is generated and sent to the target user.

[0050] The product recommendation device 100 provided in this embodiment of the invention searches for similar users among other users based on the user similarity between the target user and other users, and determines the initial product recommendation list for the target user based on the product preference information of different similar users. This ensures that the recommended products are not limited to the target user's preferred products, and can effectively cover products with different types or new products when the product preference information of similar users involves different types of products. In addition, by further filtering the obtained initial product recommendation list based on the target user's behavioral information, the accuracy and diversity of the generated target product recommendation list can be effectively improved.

[0051] Please see Figure 3 The third aspect of the present invention provides an electronic device 200, including a memory 22, a processor 21, and a computer program stored in the memory 22 and executable on the processor 21. When the processor 21 executes the computer program, it implements the product recommendation method described in any embodiment of the first aspect.

[0052] For example, the computer program may be divided into one or more modules / units, which are stored in the memory 22 and executed by the processor 21 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device 200.

[0053] The electronic device 200 may include, but is not limited to, a processor 21 and a memory 22. Those skilled in the art will understand that the schematic diagram is merely an example of the electronic device 200 and does not constitute a limitation on the electronic device 200. It may include more or fewer components than illustrated, or combine certain components, or different components. For example, the electronic device 200 may also include input / output devices, network access devices, buses, etc.

[0054] The processor 21 can be a central processing unit (CPU), or 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 the processor 21 can be any conventional processor 21. The processor 21 is the control center of the electronic device 200, connecting various parts of the electronic device 200 via various interfaces and lines.

[0055] The memory 22 can be used to store the computer programs and / or modules. The processor 21 implements various functions of the electronic device 200 by running or executing the computer programs and / or modules stored in the memory 22 and calling the data stored in the memory 22. The memory 22 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 22 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0056] A fourth aspect of the present invention provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the product recommendation method described in any embodiment of the first aspect.

[0057] A fifth aspect of the present invention provides a computer program product, including a computer program / instructions, wherein when the computer program / instructions are executed by a processor, they implement the steps of the product recommendation method described in any embodiment of the first aspect.

[0058] Wherein, if the modules / units integrated in the electronic device 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, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0059] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A product recommendation method, characterized in that, include: Based on the user similarity between the target user and the other users, a similar user search is performed on the other users to obtain a preset number of similar users; wherein, the other users are the set of users other than the target user; Based on the user similarity and the product preference information of each similar user, an initial product recommendation list for the target user is obtained; Based on the preset product classification information, the target classification information of each initially recommended product in the initial product recommendation list is determined; Based on the target classification information and the target user's behavior information, the initial product recommendation list is filtered to generate a target product recommendation list and sent to the target user.

2. The product recommendation method as described in claim 1, characterized in that, The step of searching for similar users among the remaining users based on the user similarity between the target user and other users to obtain a preset number of similar users specifically includes: Using the user similarity as the fitness value, a genetic algorithm is used to search for similar users among the remaining users to obtain a preset number of similar users; wherein, the user similarity is determined based on the similarity between the first feature information of the target user and the second feature information of the remaining users.

3. The product recommendation method as described in claim 2, characterized in that, The step of using the user similarity as the fitness value and employing a genetic algorithm to search for similar users among the remaining users to obtain a preset number of similar users specifically includes: Using the user similarity as the fitness value, a roulette wheel selection algorithm is used to perform a selection operation on the current user population to obtain a first optimized user population; wherein, the initial user population is a population composed of the remaining users. Perform a crossover operation on the first optimized user population to obtain a second optimized user population; Based on the initial user population, a mutation operation is performed on the second optimized user population to obtain a third optimized user population; Using the third optimized user population as the current user population, repeat the aforementioned steps until the fitness value of the current user population converges; based on the fitness value of the current user population, obtain a preset number of similar users.

4. The product recommendation method as described in claim 1, characterized in that, The step of obtaining the initial product recommendation list for the target user based on the user similarity and the product preference information of each similar user specifically includes: Based on the product preference information, determine the first demand rating of each similar user for each product; Based on the user similarity, the first demand score, and the average product demand score of the target user, a second demand score for each product is obtained for the target user. Based on the first demand score and the second demand score, target similar users related to the target user are determined from each of the similar users; Based on the second demand rating of the target similar users and the product association matrix of the target users, the initial product recommendation list is obtained; wherein, the product association matrix is ​​used to represent the actual demand rating of the target users for each product.

5. The product recommendation method as described in claim 4, characterized in that, The initial product recommendation list is obtained based on the second demand score of the target similar users and the product association matrix of the target users, specifically including: Based on the sparsity of the product association matrix and a preset sparsity threshold, the user type of the target user is determined; wherein, the sparsity is determined based on the number of products in the product association matrix that do not have the actual demand rating. When the user type is a new user, the initial product recommendation list is determined based on the first initial recommended product whose second demand score of the target similar user is greater than the preset demand score threshold. When the user type is not a new user, the initial product recommendation list is determined based on the first initial recommended product and the products with clearly defined needs whose actual need score is greater than the need score threshold.

6. The product recommendation method as described in claim 1, characterized in that, The step of filtering the initial product recommendation list based on the target classification information and the target user's behavior information, generating a target product recommendation list, and sending it to the target user specifically includes: Based on the behavioral information, the number of times the target user exhibits product demand behavior for each of the target categorization information is obtained; The initial product recommendation list is filtered according to the preset target number of recommended products and the number of times the product demand behavior occurs, and the target product recommendation list is generated and sent to the target user.

7. A product recommendation device, characterized in that, include: The similar user acquisition module is used to perform a similar user search on the remaining users based on the user similarity between the target user and the remaining users, and obtain a preset number of similar users; wherein, the remaining users are a set of users other than the target user; The initial product recommendation list acquisition module is used to obtain the initial product recommendation list of the target user based on the product preference information of each of the similar users; The product classification module is used to determine the target classification information of each initially recommended product in the initial product recommendation list based on preset product classification information; The product recommendation module is used to filter the initial product recommendation list based on the target classification information and the target user's behavior information, generate a target product recommendation list, and send it to the target user.

8. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the product recommendation method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the product recommendation method according to any one of claims 1 to 6.

10. A computer program product, characterized in that, Includes a computer program / instructions that, when executed by a processor, implement the steps of the product recommendation method according to any one of claims 1 to 6.

Images