Product recommendation method, device, storage medium and computer equipment

Abstract

The present application provides a product recommendation method, device, storage medium and computer equipment. When recommending products, the product exposure data of each product in each cargo pool at the current moment can be obtained, and then the product static score and support strength of each product can be determined based on the product exposure data. Then, the products in each cargo pool can be screened based on the product static score and support strength of each product in each cargo pool to obtain multiple recalled products, and the support score can be calculated based on the product static score and support strength of each recalled product. Finally, the support score of each recalled product can be sorted, and a recalled product support queue can be generated based on the sorting result and the preset support number, so that each recalled product can be recommended to the user based on the recalled product support queue. In this way, while ensuring the diversity of recalled products, recalled products with higher support scores can be recommended to the user's browsing page, so as to improve the recommendation effect of products on the e-commerce platform.

Description

Technical Field

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

[0002] With the rapid development of e-commerce, the number of various products on e-commerce platforms is huge. Faced with overloaded product information, users can search for products they are interested in through keywords and browse them. When the e-commerce platform receives the keywords entered by the user, it can recommend related products to the user's browsing page through recall, rough sorting, fine sorting, and re-sorting.

[0003] However, due to the single recall rules of the existing recall process, the recalled products that enter the sorting are highly similar. Therefore, when users search for a certain type of product, the products recommended to the user's browsing page by the e-commerce platform are highly similar, which can easily cause visual fatigue to the user and reduce the product recommendation effect of the e-commerce platform. Summary of the Invention

[0004] The purpose of this application is to solve at least one of the above-mentioned technical defects, especially the technical defect that the recall rules of the existing recall link in the prior art are single, resulting in the recalled products entering the sorting being highly similar, thereby reducing the product recommendation effect of the e-commerce platform.

[0005] This application provides a product recommendation method, which includes:

[0006] Obtain product exposure data for each product in each product pool at the current moment, and determine the product static score and support intensity for each product based on the product exposure data;

[0007] The products in each pool are screened based on their static scores and support levels to obtain multiple recalled products.

[0008] Calculate the support score for each recalled product based on its static score and support strength;

[0009] The support scores of the recalled products are sorted in descending order, and a recalled product support queue is generated according to the sorting result and a preset support number, so as to recommend the recalled products to the user based on the recalled product support queue.

[0010] Optionally, the product exposure data includes click volume, purchase volume, and exposure volume; and determining the product static score of each product based on each product exposure data includes:

[0011] For each product, an exposure quality score is calculated based on the number of clicks and exposures of the product, and a conversion rate score is calculated based on the number of purchases and exposures of the product. The product static score of the product is determined based on the exposure quality score and the conversion rate score.

[0012] Optionally, the calculation formula for the product static score is:

[0013]

[0014] In the formula, b represents the static score of the product; represents the exposure quality score, where Indicates the number of clicks, Indicates exposure; represents the conversion rate score, where Indicates the purchase quantity.

[0015] Optionally, determining the support intensity for each product based on each product exposure data includes:

[0016] For each product, obtain the product pool and the current pool exposure data of the product pool;

[0017] Determine the current support level corresponding to the cargo pool exposure data according to a preset level threshold table;

[0018] Obtain the historical support levels of the commodity pool at multiple historical moments, and determine the support strength of the commodity pool according to each historical support level and the current support level as the support strength of the commodity.

[0019] Optionally, the products in each pool are screened based on the static product score and support strength of each product in each pool to obtain multiple recalled products, including:

[0020] Sort the static scores of each commodity in each pool from high to low to obtain the static score ranking results of each pool;

[0021] Select multiple products corresponding to the preset number of filters from the corresponding product pool in the order of the static sorting results;

[0022] Among the screened commodities, commodities with support strength less than or equal to zero are eliminated to obtain multiple recalled commodities.

[0023] Optionally, the calculation of the support score for each recalled product based on the product static score and support intensity of each recalled product includes:

[0024] Select the minimum and maximum static scores of each recalled product;

[0025] For each recalled product, the support score of the recalled product is calculated based on the product static score and support strength of the recalled product, as well as the minimum and maximum product static scores among all recalled products.

[0026] Optionally, the calculation formula for the support score is:

[0027]

[0028] In the formula, a represents the support score; b represents the static score of the product; Indicates the maximum static score of each recalled product; Indicates the minimum static score of each recalled product; Indicates the level of support.

[0029] This application also provides a product recommendation device, comprising:

[0030] The data acquisition module is used to obtain the product exposure data of each product in each product pool at the current moment, and determine the product static score and support intensity of each product based on the product exposure data;

[0031] The product recall module is used to screen the products in each pool based on the static product score and support level of each product in each pool to obtain multiple recalled products;

[0032] A support score calculation module is used to calculate the support score of each recalled product based on its static score and support intensity;

[0033] The product recommendation module is used to sort the support scores of each recalled product in descending order, and generate a recalled product support queue according to the sorting results and a preset support number, so as to recommend each recalled product to the user based on the recalled product support queue.

[0034] The present application also provides a storage medium, in which computer-readable instructions are stored. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the product recommendation method as described in any of the above embodiments.

[0035] The present application also provides a computer device, comprising: one or more processors, and a memory;

[0036] The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the one or more processors, the steps of the product recommendation method described in any one of the above embodiments are performed.

[0037] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0038] The present application provides a product recommendation method, device, storage medium and computer equipment. When recommending products, the product exposure data of each product in each cargo pool at the current moment can be obtained, and the product static score and support strength of each product can be determined based on the exposure data of each product. The exposure status of each product in each cargo pool can be intuitively understood through the product static score and support strength. Then, the products in each cargo pool can be screened based on the product static score and support strength of each product in each cargo pool to obtain multiple recalled products. In this way, the products can be screened from different dimensions to ensure that the recalled products obtained after screening are diverse, and the calculation amount of the subsequent process can be reduced, thereby improving the calculation efficiency. After obtaining multiple recalled products, each recalled product can also be screened. The static score of the product and the support strength are calculated to obtain the support score of each recalled product. The support score here can represent the quality of the recalled product. This application comprehensively calculates the data of different dimensions of each recalled product to prevent a single dimension from having a greater impact on the recalled product, resulting in a lower accuracy of the product support score. Finally, the support scores of each recalled product can be sorted in order from high to low, and a recalled product support queue can be generated according to the sorting results and the preset number of supports, so that each recalled product can be recommended to the user based on the recalled product support queue. In this way, while ensuring the diversity of the recalled products, the recalled products with higher support scores can be recommended to the user's browsing page, so as to improve the recommendation effect of products on the e-commerce platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0040] Figure 1 A schematic flow chart of a product recommendation method provided in an embodiment of the present application;

[0041] Figure 2 A schematic flow chart of a method for determining the intensity of product support provided in an embodiment of the present application;

[0042] Figure 3 A schematic flow chart of a method for screening recalled products provided in an embodiment of the present application;

[0043] Figure 4 A schematic diagram of the structure of a product recommendation device provided in an embodiment of the present application;

[0044] Figure 5A schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0046] With the rapid development of e-commerce, the number of various products on e-commerce platforms is huge. Faced with overloaded product information, users can search for products they are interested in through keywords and browse them. When the e-commerce platform receives the keywords entered by the user, it can recommend related products to the user's browsing page through recall, rough sorting, fine sorting, and re-sorting.

[0047] However, due to the single recall rules of the existing recall process, the recalled products that enter the sorting are highly similar. Therefore, when users search for a certain type of product, the products recommended to the user's browsing page by the e-commerce platform are highly similar, which can easily cause visual fatigue to the user and reduce the product recommendation effect of the e-commerce platform.

[0048] Based on this, this application proposes the following technical solutions, please refer to the following for details:

[0049] In one embodiment, Figure 1 As shown, Figure 1 This is a schematic flow chart of a product recommendation method provided in an embodiment of the present application. This application provides a product recommendation method, which specifically includes the following:

[0050] S110: Obtain product exposure data of each product in each product pool at the current moment, and determine the product static score and support strength of each product based on the product exposure data.

[0051] In this step, when the e-commerce platform detects that there is any user browsing the product page of a product in the front end, it can obtain some interactive information generated by the user's interaction with the product, form product exposure data and store it in the product database. Then the e-commerce platform can obtain the product exposure data of each product in each cargo pool at the current moment, and determine the static score and support strength of each product based on the exposure data of each product.

[0052] It can be understood that the cargo pool of the present application is composed of multiple commodities of the same commodity category. The commodities in different cargo pools correspond to different commodity categories. When a commodity is marked with multiple commodity category labels, the commodity can exist in multiple cargo pools at the same time; and the commodity static score is an indicator for measuring the popularity of the commodity, which can be calculated based on the commodity exposure data of the commodity to represent the popularity of the commodity. The higher the static score of the commodity, the more popular the commodity is and the more favored it is by users; the support strength of the commodity refers to the strength of regulating the support traffic for the commodity, which represents the investment level of the e-commerce platform in the commodity. The greater the support strength, the higher the support traffic of the commodity and the greater the exposure.

[0053] Furthermore, the e-commerce platform can obtain the product exposure data of each product in each cargo pool at the current moment by setting a scheduled task. The duration of the current moment here can be the cycle duration in the scheduled task, that is, the e-commerce platform can obtain the product exposure data generated in the time period with the time point of the last acquisition of product exposure data as the start time and the time point of this acquisition of product exposure data as the end time. The duration of the current moment can also be set separately according to the actual needs and accuracy of adjusting the support intensity of the product, such as one week, two weeks or one month.

[0054] S120: Filter the products in each pool based on the static product score and support strength of each product in each pool to obtain multiple recalled products.

[0055] In this step, after determining the static score and support strength of each commodity in step S110, for each cargo pool, each commodity can be screened based on the static score and support strength of each commodity in the cargo pool, and the filtered commodities in each cargo pool are used as recalled commodities.

[0056] It is understandable that the static score of a product is an indicator that reflects the popularity of the product, and the support strength is the degree of investment of the e-commerce platform in the product. By comprehensively considering these two factors to recall products, more popular products and products with higher investment levels can be covered, and the accuracy and coverage of the recall results can be improved, thereby improving the recommendation effect of the recalled products. At the same time, using the static score of the product and the support strength as screening conditions can achieve rapid screening of massive products, narrow the recall scope, and thus improve the efficiency of the recall.

[0057] Furthermore, when any product is marked with multiple product categories and exists in multiple cargo pools at the same time, after the product is recalled, the cargo pool exposure data of the multiple cargo pools corresponding to the recalled product can be obtained and compared, and the cargo pool with the highest cargo pool exposure data in the comparison results can be selected as the cargo pool corresponding to the recalled product, and the product static score and support strength of the recalled product in the cargo pool can be used as the final product static score and support strength.

[0058] S130: Calculate the support score of each recalled product based on the static product score and support strength of each recalled product.

[0059] In this step, after multiple recalled products are screened in step S120, the product static score and support strength of each recalled product can be obtained, and the support score of each recalled product can be calculated based on the product static score and support strength of each recalled product.

[0060] It is understandable that the support score here is calculated through two aspects: the static score of the product and the support strength. The data on two different dimensions, namely the popularity of the product and the degree of investment of the e-commerce platform in the product, can be mapped to the same dimension. Then, a comprehensive evaluation of each recalled product can be conducted based on the support score to improve the recall effect and efficiency of the product.

[0061] S140: Sort the support scores of the recalled products in descending order, and generate a recalled product support queue according to the sorting results and a preset support number, so as to recommend each recalled product to the user based on the recalled product support queue.

[0062] In this step, after calculating the support score of each recalled product through step S130, the support score of each recalled product can be sorted in order from high to low to obtain the sorting result of the support score. Then, a recalled product support queue can be generated based on the sorting result and the preset support number, so that the e-commerce platform can recommend each recalled product to the user according to the recalled product support queue.

[0063] Specifically, when generating a recalled product support queue, you can first determine the preset number of supports, and then select the recalled products in order according to the order in the sorting results until the number of selected recalled products is equal to the preset number of supports. Finally, you can arrange the selected recalled products in order from high to low according to the support scores to obtain a recalled product support queue.

[0064] Furthermore, after generating a recalled product support queue, the backend of the e-commerce platform can allocate the traffic share of the recalled products according to the support score of each recalled product in the recalled product support queue, and recommend each recalled product to the user's browsing page based on the adjusted traffic share. It can be understood that when the support score of the recalled product is higher, the traffic share allocated to the recalled product is higher, and the probability of recommending it to the user's browsing page is greater, thereby increasing the exposure rate of the recalled product. Conversely, when the support score of the recalled product is lower, the traffic share allocated to the recalled product is lower, and the probability of recommending it to the user's browsing page is smaller, thereby controlling the exposure rate of the recalled product.

[0065] In addition, the preset number of supports in this application can be set according to the actual needs and support strategies of the e-commerce platform. When the preset number of supports is greater, the overall exposure effect is improved, but the average traffic proportion allocated to each recalled product is smaller, and the exposure effect for a single recalled product will decrease. When the preset number of supports is smaller, the average traffic proportion allocated to each recalled product by the e-commerce platform increases, and the exposure effect for a single recalled product will be significantly improved.

[0066] In the above embodiment, when recommending products, the product exposure data of each product in each cargo pool at the current moment can be obtained, and the product static score and support strength of each product can be determined based on the exposure data of each product. The exposure status of each product in each cargo pool can be intuitively understood through the product static score and support strength. Then, the products in each cargo pool can be screened based on the product static score and support strength of each product in each cargo pool to obtain multiple recalled products. In this way, the products can be screened from different dimensions to ensure that the recalled products obtained after screening are diverse, and the amount of calculation in the subsequent process can be reduced, thereby improving the operation efficiency. After obtaining multiple recalled products, the product static score and support strength of each recalled product can also be Calculations are performed to obtain the support score of each recalled product. The support score here can represent the quality level of the recalled product. This application performs comprehensive calculations on the data of different dimensions of each recalled product to prevent a single dimension from having a greater impact on the recalled product, resulting in a lower accuracy of the product support score. Finally, the support scores of each recalled product can be sorted in descending order, and a recalled product support queue can be generated based on the sorting results and the preset number of supports, so that each recalled product can be recommended to the user based on the recalled product support queue. In this way, while ensuring the diversity of the recalled products, recalled products with higher support scores can be recommended to the user's browsing page to improve the recommendation effect of products on the e-commerce platform.

[0067] In one embodiment, the product exposure data in step S110 may include click volume, purchase volume, and exposure volume; wherein determining the product static score of each product based on each product exposure data may include:

[0068] S111: For each product, an exposure quality score is calculated based on the number of clicks and exposures of the product, and a conversion rate score is calculated based on the number of purchases and exposures of the product, and a static product score of the product is determined based on the exposure quality score and the conversion rate score.

[0069] In this embodiment, after obtaining the product exposure data of each product, the number of clicks, purchases and exposures in the exposure data of each product can be obtained first, and then the exposure quality score can be calculated based on the number of clicks and exposures of the product, and the conversion rate score can be calculated based on the purchases and exposures of the product. Finally, the product static score of the product can be calculated based on the exposure quality score and the conversion rate score.

[0070] It's understandable that clicks refer to the number of times a user clicks through a browsing page to access the product's details page during the current moment; purchases refer to the number of times a user purchases the product through a browsing page during the current moment; and exposure refers to the number of times a product appears on a user's browsing page during the current moment. Furthermore, the exposure quality score reflects the product's exposure and quality; a higher score indicates better exposure and more effective recommendations to users. The conversion rate score reflects the product's conversion effectiveness and profitability; a higher score indicates better conversion, potentially generating more revenue and profit for the e-commerce platform.

[0071] Furthermore, before calculating the click volume, purchase volume, and impression volume for each product, the click volume, purchase volume, and impression volume can be normalized. Data from different dimensions has different value ranges and influences the recall results to varying degrees. Therefore, to prevent data from certain dimensions from having a disproportionate impact on the recall results, the data from each dimension can be normalized and mapped to the [0, 1] interval for comprehensive calculation.

[0072] In one embodiment, the calculation formula for the product static score in step S111 is:

[0073]

[0074] In the formula, b represents the static score of the product; represents the exposure quality score, where Indicates the number of clicks, Indicates exposure; represents the conversion rate score, where Indicates the purchase quantity.

[0075] In this embodiment, Indicates the exposure quality score of the product, where It can prevent the numerator from being 0, 1000 and map the score to the interval [0, 1000]. It can balance the relationship between exposure and click volume, and adding 1 at the end can avoid the exposure quality score being 0; Represents the conversion rate score of the product, where It can prevent the numerator from being 0, It can balance the relationship between exposure and conversion, and adding 1 at the end can avoid a conversion rate score of 0.

[0076] Furthermore, the use of a logarithmic function in the calculation formula for a product's static score to calculate the exposure quality score and conversion rate score can limit the range of the exposure quality score and conversion rate score to a smaller interval, making the difference between the exposure quality score and the conversion rate score more obvious. Since the logarithmic function also has a smoothing characteristic, when the score changes slightly, the logarithm can make the score change smoother.

[0077] In one embodiment, Figure 2 As shown, Figure 2 A schematic flow chart of a method for determining the intensity of product support provided in an embodiment of the present application; Figure 2 In step S110, determining the support intensity for each product based on the exposure data of each product may include:

[0078] S112: For each product, obtain the product pool and the current pool exposure data of the product pool.

[0079] S113: Determine the current support level corresponding to the cargo pool exposure data according to a preset level threshold table.

[0080] S114: Obtain the historical support levels of the commodity pool at multiple historical moments, and determine the support strength of the commodity pool based on each historical support level and the current support level as the support strength of the commodity.

[0081] In this embodiment, when calculating the support strength for each commodity, for each commodity, the cargo pool in which the commodity is located and the cargo pool exposure data of the cargo pool at the current moment can be obtained first, and then the cargo pool exposure data can be compared with the threshold of each level in the preset level threshold table to determine the current support level of the cargo pool based on the comparison result, and finally the historical support levels of the cargo pool at multiple historical moments can be obtained, and the support strength of the cargo pool can be determined based on each historical support level and the current support level as the support strength of the commodity.

[0082] It is understandable that when obtaining the preset level threshold table, a comprehensive evaluation of the product pool can be conducted based on factors such as different business scenarios, product attributes, and market environment. Data such as click volume, purchase volume, and exposure volume can be used as indicators to set corresponding indicator thresholds for each support level to form a preset level threshold table. When there are changes in factors, the indicator thresholds in the preset level threshold table can also be adjusted. In addition, the multiple historical moments in this application can refer to multiple historical moments that are sequentially pushed forward from the current moment, and there is no limit on the number of historical moments obtained.

[0083] Specifically, the present application can sum up the product exposure data of each product in the product pool at the current moment to obtain the product pool exposure data. When summing the product exposure data of each product, you can first determine the indicator data that needs to be counted, such as click volume, purchase volume, and exposure volume, etc. Then, you can clean each indicator data to remove invalid data such as machine exposure and repeated exposure, so as to reduce the data that needs to be integrated and reduce the load pressure on the server when processing data, thereby improving data processing capabilities and efficiency.

[0084] Furthermore, this application can use the PID calculation formula to adjust the current support level to obtain the support strength of the product. The specific calculation formula is as follows:

[0085]

[0086] Where, is the current support level of the cargo pool at the current moment, It is the historical support level of the commodity pool at the previous moment. It is the historical support level of the cargo pool at the second-to-last moment, Kp is the deviation term coefficient, Ki is the integral term coefficient, and Kd is the differential term coefficient.

[0087] Among them, Kp, Ki, and Kd are three parameters in the PID calculation formula, which represent the coefficients of the proportional term, integral term, and differential term, respectively. They correspond to the deviation, integral, and differential amounts, respectively. In practical applications, PID parameters can be selected in combination with specific application scenarios and actual control requirements. In this application, Kp can be 1, Ki can be 3, and Kd can be 0.5.

[0088] In one embodiment, Figure 3 As shown, Figure 3 A schematic flow chart of a method for screening recalled products provided in an embodiment of the present application; Figure 3 In step S120, the products in each pool are screened based on the static product score and support strength of each product in each pool to obtain multiple recalled products, which may include:

[0089] S121: Sort the static scores of the commodities in each commodity pool from high to low to obtain the static score sorting results of each commodity pool.

[0090] S122: Selecting a plurality of commodities corresponding to a preset number of screenings from the corresponding commodity pool in sequence according to the order of the static sorting results.

[0091] S123: Eliminate the products whose support strength is less than or equal to zero from the screened products to obtain multiple recalled products.

[0092] In this embodiment, when screening the commodities in each cargo pool, the static scores of the commodities in each cargo pool can be first sorted from high to low to obtain the static score sorting results of each cargo pool. Then, multiple commodities corresponding to the preset screening number can be selected from the corresponding cargo pool in sequence according to the order of the static score sorting results, and the commodities with support strength less than or equal to zero in the screened commodities can be eliminated to obtain multiple recalled commodities.

[0093] Specifically, for each cargo pool, the static score of each commodity in the cargo pool can be obtained, so that the static score of each commodity can be sorted from high to low to obtain the static score sorting result. Then, the preset number of filters can be determined based on the business needs and recommendation effect of the cargo pool. When the popularity of the cargo pool is higher, the number of filters for the cargo pool can be appropriately increased to ensure that the screening results can cover more popular and high-quality commodities. At the same time, the number of filters also needs to be controlled to prevent excessive number of filters from reducing computing efficiency. After determining the preset number of filters, the commodities in the cargo pool can be selected in sequence according to the order in the static score sorting result until the number of selected commodities is equal to the preset number of filters.

[0094] Furthermore, after screening the products in each pool, the products with support strength less than or equal to zero can be eliminated from the screened products to obtain multiple recalled products. Here, the support strength of the product is less than or equal to zero, indicating that the product exposure data of the product has reached the standard or is even in a state of overexposure. Therefore, traffic support can be omitted for this type of product, and this type of product can be eliminated from the recalled products.

[0095] In one embodiment, step S130 calculates the support score of each recalled product based on the product static score and support strength of each recalled product, including:

[0096] S131: Select the minimum product static score and the maximum product static score among the product static scores of each recalled product.

[0097] S132: For each recalled product, calculate the support score of the recalled product based on the product static score and support strength of the recalled product, as well as the minimum product static score and the maximum product static score among all recalled products.

[0098] In this embodiment, when calculating the support score of each recalled product, the product static score of each recalled product can be sorted, and the maximum value in the sorting result can be selected as the maximum product static score, and the minimum value in the sorting result can be selected as the minimum product static score. When calculating the support score of each recalled product separately, the support score of the recalled product can be calculated based on the product static score and support strength of the recalled product, as well as the minimum product static score and the maximum product static score.

[0099] In one embodiment, the calculation formula for the support score in step S132 is:

[0100]

[0101] In the formula, a represents the support score; b represents the static score of the product; Indicates the maximum static score of each recalled product; Indicates the minimum static score of each recalled product; Indicates the level of support.

[0102] In this embodiment, for each recalled product, the product static score and support strength of the recalled product can be obtained first, and then the support score of the recalled product can be calculated based on the product static score and support strength, as well as the ratio between the maximum product static score and the minimum product static score.

[0103] The ratio between the maximum and minimum static scores of each recalled product can be used to measure the degree of difference between the maximum and minimum static scores of each recalled product. Specifically, a larger ratio indicates a greater range of variation in the data and a more pronounced difference; conversely, a smaller ratio indicates a smaller range of variation in the data and a less pronounced difference.

[0104] The following describes a product recommendation device provided in an embodiment of the present application. The product recommendation device described below and the product recommendation method described above can be referenced to each other.

[0105] In one embodiment, Figure 4 As shown, Figure 4 This is a schematic diagram of the structure of a product recommendation device provided in an embodiment of the present application. The present application also provides a product recommendation device, including a data acquisition module 210, a product recall module 220, a support score calculation module 230, and a product recommendation module 240, specifically including the following:

[0106] The data acquisition module 210 is used to obtain the product exposure data of each product in each product pool at the current moment, and determine the product static score and support strength of each product based on the product exposure data.

[0107] The product recall module 220 is used to screen the products in each pool based on the static product score and support strength of each product in each pool to obtain multiple recalled products.

[0108] The support score calculation module 230 is used to calculate the support score of each recalled product based on the product static score and support strength of each recalled product.

[0109] The product recommendation module 240 is used to sort the support scores of each recalled product in descending order, and generate a recalled product support queue according to the sorting results and a preset support number, so as to recommend each recalled product to the user based on the recalled product support queue.

[0110] In the above embodiment, when recommending products, the product exposure data of each product in each cargo pool at the current moment can be obtained, and the product static score and support strength of each product can be determined based on the exposure data of each product. The exposure status of each product in each cargo pool can be intuitively understood through the product static score and support strength. Then, the products in each cargo pool can be screened based on the product static score and support strength of each product in each cargo pool to obtain multiple recalled products. In this way, the products can be screened from different dimensions to ensure that the recalled products obtained after screening are diverse, and the amount of calculation in the subsequent process can be reduced, thereby improving the operation efficiency. After obtaining multiple recalled products, the product static score and support strength of each recalled product can also be Calculations are performed to obtain the support score of each recalled product. The support score here can represent the quality level of the recalled product. This application performs comprehensive calculations on the data of different dimensions of each recalled product to prevent a single dimension from having a greater impact on the recalled product, resulting in a lower accuracy of the product support score. Finally, the support scores of each recalled product can be sorted in descending order, and a recalled product support queue can be generated based on the sorting results and the preset number of supports, so that each recalled product can be recommended to the user based on the recalled product support queue. In this way, while ensuring the diversity of the recalled products, recalled products with higher support scores can be recommended to the user's browsing page to improve the recommendation effect of products on the e-commerce platform.

[0111] In one embodiment, the data acquisition module 210 may include:

[0112] The static score calculation submodule is used to calculate the exposure quality score of each product based on the number of clicks and exposures of the product, and to calculate the conversion rate score based on the purchase volume and exposure of the product, and to determine the product static score based on the exposure quality score and conversion rate score.

[0113] In one embodiment, the data acquisition module 210 may further include:

[0114] The cargo pool exposure data acquisition submodule is used to obtain the cargo pool where the product is located and the cargo pool exposure data of the cargo pool at the current moment for each product.

[0115] The current support level determination submodule is used to determine the current support level corresponding to the cargo pool exposure data according to the preset level threshold table.

[0116] The support strength calculation submodule is used to obtain the historical support levels of the cargo pool at multiple historical moments, and determine the support strength of the cargo pool based on each historical support level and the current support level as the support strength of the product.

[0117] In one embodiment, the product recall module 220 may include:

[0118] The static score sorting submodule is used to sort the static scores of each commodity in each cargo pool from high to low to obtain the static score sorting results of each cargo pool.

[0119] The product screening submodule is used to select multiple products corresponding to the preset screening number from the corresponding cargo pool in the order of each static sorting result.

[0120] The product elimination submodule is used to eliminate the products whose support strength is less than or equal to zero from the screened products, thereby obtaining multiple recalled products.

[0121] In one embodiment, the support score calculation module 230 may include:

[0122] The static score selection submodule is used to select the minimum static score and the maximum static score of each recalled product.

[0123] The support score calculation submodule is used to calculate the support score of each recalled product based on the product static score and support strength of the recalled product, as well as the minimum and maximum product static scores among all recalled products.

[0124] In one embodiment, the present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the product recommendation method as described in any of the above embodiments.

[0125] In one embodiment, the present application also provides a computer device having computer-readable instructions stored therein. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the product recommendation method as described in any one of the above embodiments.

[0126] Schematically, as Figure 5 As shown, Figure 5 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. The computer device 300 can be provided as a server. Figure 5Computer device 300 includes a processing component 302, which further includes one or more processors, and a memory resource represented by memory 301 for storing instructions executable by processing component 302, such as an application. The application stored in memory 301 may include one or more modules, each corresponding to a set of instructions. In addition, processing component 302 is configured to execute the instructions to perform the product recommendation method of any of the above-mentioned embodiments.

[0127] The computer device 300 may further include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate based on an operating system stored in the memory 301, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or the like.

[0128] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0129] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0130] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referenced to each other.

[0131] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A product recommendation method, characterized in that: The method comprises: Obtain product exposure data for each product in each product pool at the current moment, and determine the product static score and support intensity for each product based on the product exposure data; The products in each pool are screened based on their static scores and support levels to obtain multiple recalled products. Calculate the support score for each recalled product based on its static product score and support strength; Sort the support scores of the recalled products in descending order, and generate a recalled product support queue based on the sorting results and a preset number of supported products, so as to recommend the recalled products to the user based on the recalled product support queue; The products in each pool are screened based on the static product score and support strength of each product in each pool to obtain multiple recalled products, including: Sort the static scores of each commodity in each pool from high to low to obtain the static score ranking results of each pool; Select multiple products corresponding to the preset number of filters from the corresponding product pool in the order of the static sorting results; Eliminate products with support levels less than or equal to zero from the screened products to obtain multiple recalled products; The support score for each recalled product is calculated based on the product static score and support strength of each recalled product, including: Select the minimum and maximum static scores of each recalled product; For each recalled product, the support score is calculated based on the product's static score and support strength, as well as the minimum and maximum static scores among all recalled products. The calculation formula for the support score is: ； In the formula, a represents the support score; b represents the static score of the product; Indicates the maximum static score of each recalled product; Indicates the minimum static score of each recalled product; Indicates the level of support.

2. The product recommendation method according to claim 1, characterized in that: The product exposure data includes click volume, purchase volume, and exposure volume; and determining the product static score of each product based on each product exposure data includes: For each product, an exposure quality score is calculated based on the number of clicks and exposures of the product, and a conversion rate score is calculated based on the number of purchases and exposures of the product. The product static score of the product is determined based on the exposure quality score and the conversion rate score.

3. The product recommendation method according to claim 2, characterized in that: The calculation formula for the static score of the product is: ； In the formula, b represents the static score of the product; represents the exposure quality score, where Indicates the number of clicks, Indicates exposure; represents the conversion rate score, where Indicates the purchase quantity.

4. The product recommendation method according to claim 1, wherein: The determination of the support intensity for each product based on each product exposure data includes: For each product, obtain the product pool and the current pool exposure data of the product pool; Determine the current support level corresponding to the cargo pool exposure data according to a preset level threshold table; Obtain the historical support levels of the commodity pool at multiple historical moments, and determine the support strength of the commodity pool according to each historical support level and the current support level as the support strength of the commodity.

5. A product recommendation device, characterized in that: include: The data acquisition module is used to obtain the product exposure data of each product in each product pool at the current moment, and determine the product static score and support intensity of each product based on the product exposure data; The product recall module is used to screen the products in each pool based on the static product score and support level of each product in each pool to obtain multiple recalled products; A support score calculation module is used to calculate the support score of each recalled product based on its static score and support intensity; A product recommendation module is used to sort the support scores of each recalled product from high to low, and generate a recalled product support queue based on the sorting results and a preset support number, so as to recommend each recalled product to the user based on the recalled product support queue; The product recall module includes: Sort the static scores of each commodity in each pool from high to low to obtain the static score ranking results of each pool; Select multiple products corresponding to the preset number of filters from the corresponding product pool in the order of the static sorting results; Eliminate products with support levels less than or equal to zero from the screened products to obtain multiple recalled products; The support score calculation module includes: Select the minimum and maximum static scores of each recalled product; For each recalled product, the support score is calculated based on the product's static score and support strength, as well as the minimum and maximum static scores among all recalled products. The calculation formula for the support score is: ； In the formula, a represents the support score; b represents the static score of the product; Indicates the maximum static score of each recalled product; Indicates the minimum static score of each recalled product; Indicates the level of support.

6. A storage medium, characterized in that: The storage medium stores computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to execute the steps of the product recommendation method according to any one of claims 1 to 4.

7. A computer device, characterized in that: include: one or more processors, and memory; The memory stores computer-readable instructions, which, when executed by the one or more processors, execute the steps of the product recommendation method according to any one of claims 1 to 4.

Images