Merchandise rearrangement method, device, apparatus and storage medium

Abstract

The application relates to a commodity rearrangement method, device, equipment and storage medium. Main technical solutions comprise the following steps: obtaining a user request, searching according to the user request, and obtaining a plurality of matched commodities; inputting the plurality of matched commodities into an initial arrangement model to obtain a plurality of commodities to be rearranged; obtaining an initial arrangement score of each commodity to be rearranged, obtaining a sorting factor of each commodity to be rearranged according to the initial arrangement score, obtaining a preset bottoming factor, comparing each sorting factor with the preset bottoming factor, and selecting the maximum as a rearrangement factor of each commodity to be rearranged; performing a rearrangement operation on each commodity to be rearranged according to the rearrangement factor, and displaying the rearranged result to the user. The application adopts a dynamic truncation mode to disturb the original sorting scene at the least cost, achieves the effect of the least influence on the initial arrangement result, guarantees the click rate and conversion rate, and ensures that the rearranged commodities meet the demands of the user.

Description

Technical Field

[0001] This application relates to the field of e-commerce traffic control technology, and in particular to a product rearrangement method, apparatus, device, and storage medium. Background Technology

[0002] With the rapid development of internet technology, online services have become widely used. In online services, product providers publish their products on online platforms, showcasing them to users so that users can reference product information and make purchases. However, because online platforms contain massive amounts of product data, not all of it can be displayed to users. Therefore, it is necessary to filter out a portion of the products that meet the user's specific requirements from this vast amount of data and present them to the user in a sorted manner.

[0003] In traditional implementations, user requests typically go through several stages: recall, initial ranking, and re-ranking, before products are displayed to the user. Usually, after the initial ranking, the results represent the best click-through rate and conversion rate. However, this ranking cannot be directly shown to the user because factors such as national policies, company strategies, and operational rules must be considered. Therefore, re-ranking becomes an essential step. For example, if the initial ranking has 1500 results, and some products from lower ranks are moved to higher ranks during re-ranking, it will ultimately lead to a decrease in conversion and click-through rates. Therefore, to minimize the impact on the initial ranking and ensure click-through and conversion rates, a fixed truncation parameter is usually set during re-ranking, ensuring that the re-ranking operation only processes products before the truncation parameter.

[0004] However, different user requests target different products. If a truncation parameter is preset for all products, the rearrangement operation will only process the products before the truncation parameter. This approach lacks personalization because the differences between different products can be very large. If a truncation parameter is used for all products, all products will be treated the same, ignoring the differences between them. Consequently, the rearranged products may not adequately meet the user's specific needs. Summary of the Invention

[0005] Based on this, this application provides a product rearrangement method, apparatus, device, and storage medium. By calculating the rearrangement factor of each product to be rearranged, the rearrangement operation is performed on each product based on the rearrangement factor, so as to minimize the impact on the initial ranking result and ensure that the rearranged products better meet the needs of users while ensuring click-through rate and conversion rate.

[0006] Firstly, a method for rearranging goods is provided, the method comprising:

[0007] Obtain the user request, search based on the user request, and obtain multiple matching products;

[0008] Input multiple matching products into the initial ranking model to obtain multiple products to be rearranged;

[0009] Obtain the initial ranking score of each item to be rearranged, and obtain the sorting factor of each item based on the initial ranking score;

[0010] Obtain the preset support factor, compare each sorting factor with the preset support factor, and select the largest one as the sorting factor for each product to be sorted.

[0011] The rearrangement operation is performed on each product to be rearranged based on each rearrangement factor, and the rearranged results are displayed to the user.

[0012] According to one achievable method in an embodiment of this application, the initial ranking model includes a coarse ranking model and a fine ranking model. Multiple matching products are input into the initial ranking model to obtain multiple products to be rearranged, including:

[0013] Filter multiple matching products to obtain multiple filtered products;

[0014] Input multiple filtered products into the coarse-ranking model to obtain multiple coarse-ranked products;

[0015] Input multiple coarse-ranked items into the fine-ranking model to obtain multiple items to be rearranged.

[0016] According to one feasible method in an embodiment of this application, the sorting factor of each product to be rearranged is obtained based on each initial sort score, including:

[0017] Obtain the maximum score from each initial ranking score, and calculate the ratio of the maximum score to each initial ranking score to obtain the multiple score for each item to be rearranged.

[0018] Obtain the preset limit value and preset hyperparameter value, and based on the preset limit value, preset hyperparameter value and each multiplier, obtain the sorting factor for each product to be rearranged.

[0019] According to one achievable method in the embodiments of this application, the preset hyperparameter value includes a preset control strength value and a preset protrusion degree value. Based on the preset limit value, the preset hyperparameter value, and each multiplier, the sorting factor for each product to be rearranged is obtained, including:

[0020] Based on the multiples and the preset protrusion values, the first control factor for each product to be rearranged is obtained;

[0021] The second control factor is obtained based on the preset control strength value and the preset protrusion degree value;

[0022] Based on the preset limit value, each first control factor, and the second control factor, the sorting factor for each product to be rearranged is obtained.

[0023] According to one feasible method in an embodiment of this application, the sorting factor of each product to be rearranged is obtained based on a preset limit value, each first control factor, and a second control factor, including:

[0024] The ratios of each first regulatory factor to the second regulatory factor are calculated to obtain the regulatory values ​​for each commodity to be rearranged.

[0025] The difference between the preset limit value and each control value is calculated to obtain the sorting factor of each product to be rearranged.

[0026] According to one feasible method in the embodiments of this application, the items to be rearranged are rearranged according to each rearrangement factor, and the rearranged results are displayed to the user, including:

[0027] The initial score of each item to be rearranged is multiplied by its corresponding rearrangement factor to obtain the weighted score of each item to be rearranged.

[0028] The products to be rearranged are rearranged based on their initial ranking score and corresponding weighted score, and the rearranged results are displayed to the user.

[0029] According to one feasible method in the embodiments of this application, the products to be rearranged are rearranged based on their initial ranking score and corresponding weighted score, and the rearranged results are displayed to the user, including:

[0030] The initial ranking score of each item to be rearranged is summed with its corresponding weighted score to obtain the rearrangement score of each item to be rearranged.

[0031] The products to be rearranged are rearranged in descending order of their rearrangement scores, and the results are displayed to the user.

[0032] Secondly, a commodity rearrangement apparatus is provided, the apparatus comprising:

[0033] The matching module is used to obtain user requests, search based on user requests, and obtain multiple matching products;

[0034] The initial ranking module is used to input multiple matching products into the initial ranking model to obtain multiple products to be rearranged.

[0035] The sorting factor acquisition module is used to obtain the initial sorting score of each product to be rearranged, and to obtain the sorting factor of each product to be rearranged based on the initial sorting score.

[0036] The module for obtaining rearrangement factors is used to obtain preset support factors, compare each sorting factor with the preset support factors, and select the largest one as the rearrangement factor for each product to be rearranged.

[0037] The rearrangement module is used to rearrange each product to be rearranged based on various rearrangement factors, and then display the rearranged results to the user.

[0038] Thirdly, a computer device is provided, comprising:

[0039] At least one processor; and

[0040] A memory that is communicatively connected to at least one processor; wherein,

[0041] The memory stores computer instructions that can be executed by at least one processor to enable the at least one processor to perform the methods involved in the first aspect above.

[0042] Fourthly, a computer-readable storage medium is provided, having stored thereon computer instructions, characterized in that the computer instructions are used to cause a computer to perform the methods involved in the first aspect above.

[0043] According to the technical content provided in the embodiments of this application, by obtaining user requests, searching according to user requests, and obtaining multiple matching products; inputting multiple matching products into an initial ranking model to obtain multiple products to be re-ranked; obtaining the initial ranking score of each product to be re-ranked, and obtaining the ranking factor of each product to be re-ranked based on the initial ranking score; obtaining a preset support factor, comparing each ranking factor with the preset support factor, and selecting the largest one as the re-ranking factor of each product to be re-ranked; performing a re-ranking operation on each product to be re-ranked based on each re-ranking factor, and displaying the re-ranked result to the user. The above operation adopts a dynamic truncation method, by calculating the re-ranking factor of each product to be re-ranked, performing a re-ranking operation on each product to be re-ranked based on the re-ranking factor, and displaying the re-ranked result to the user, in order to achieve the effect of disrupting the original ranking scenario at the lowest possible cost, and while ensuring click-through rate and conversion rate, ensuring that the re-ranked products better meet the user's needs. Attached Figure Description

[0044] Figure 1 This is a diagram illustrating the application environment of a product rearrangement method in one embodiment.

[0045] Figure 2 This is a flowchart illustrating a product rearrangement method in one embodiment;

[0046] Figure 3 This is a schematic diagram of a preferred process for a product rearrangement method in one embodiment;

[0047] Figure 4 This is a structural block diagram of a product rearrangement device in one embodiment;

[0048] Figure 5This is a schematic structural diagram of a computer device in one embodiment. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0050] This application provides a product rearrangement method that can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. Server 104 receives user requests, searches based on the requests, and obtains multiple matching products; inputs these matching products into an initial ranking model to obtain multiple products to be rearranged; obtains the initial ranking score for each product to be rearranged, and calculates the ranking factor for each product based on the initial ranking score; obtains a preset safety net factor, compares each ranking factor with the preset safety net factor, and selects the largest one as the rearrangement factor for each product to be rearranged; performs a rearrangement operation on each product to be rearranged based on the rearrangement factors, and displays the rearranged results to the user. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, and tablets, etc., and server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0051] In one embodiment, such as Figure 2 As shown, a product rearrangement method is provided, which can be applied to... Figure 1 Taking server 104 as an example, the following steps are included:

[0052] Step S201: Obtain the user request, perform a search based on the user request, and obtain multiple matching products.

[0053] Here, the server receives the product request sent by the end user and searches the online platform database based on the user's request to obtain multiple products that match the user's request. For example, if the user enters "red women's shoes under 500 yuan," the server can then search the online platform database for all products related to women's shoes.

[0054] Step S203: Input multiple matching products into the initial ranking model to obtain multiple products to be rearranged.

[0055] The initial ranking model is a pre-set model based on the specifications, colors, prices, and target audiences of different products.

[0056] Here, since the multiple matching products may not all meet the user's needs, it is necessary to input these matching products into the initial ranking model to obtain multiple products to be rearranged that meet the user's specific needs. Assuming the user inputs "red women's shoes under 500 yuan," the initial matching will search for all products related to women's shoes. Inputting these searched products into the initial ranking model will then yield multiple products to be rearranged for "red women's shoes under 500 yuan."

[0057] Step S205: Obtain the initial ranking score of each item to be rearranged, and obtain the sorting factor of each item to be rearranged based on the initial ranking score.

[0058] Here, after the initial ranking, the click-through rate and conversion rate are already relatively good. However, factors such as national policies, company strategies, and operational rules still need to be considered. For example, if everyone excludes a certain brand from a certain country at a certain time, then that brand's products will be blocked. Therefore, re-ranking becomes an essential step. This can be done by obtaining the initial ranking score of each product to be re-ranked, and then using that score to determine the ranking factor for each product. It's important to note that after inputting multiple matching products into the initial ranking model, the model will score each matching product, thus obtaining the initial ranking score for each product to be re-ranked.

[0059] Step S207: Obtain the preset support factor, compare each sorting factor with the preset support factor, and select the largest one as the sorting factor for each product to be sorted.

[0060] Among them, the preset support factor is a support parameter that is preset by the staff in advance. This is because the sorting factor of the products to be rearranged based on the initial sorting may be less than the preset support factor. However, if the sorting factor is too small, the rearrangement operation will be meaningless. Therefore, a support parameter can be preset in advance.

[0061] Here, we assume the preset safety factor is represented by `baseFactor`, and the sorting factor for the i-th item to be rearranged is represented by `sortingFactor`. i This indicates that the rearrangement factor for the i-th item to be rearranged is represented by the rearrangementFactor. i The expression for comparing the i-th sorting factor with the preset support factor and selecting the larger one as the rearrangement factor for the i-th item to be rearranged is as follows:

[0062] rearrangementFactor i =max(sortingFactor) i (baseFactor)

[0063] Similarly, the rearrangement factor of each item to be rearranged can be obtained through the above expression.

[0064] Step S209: Rearrange each item to be rearranged according to each rearrangement factor, and display the rearranged results to the user.

[0065] Here, the products to be rearranged are rearranged according to each rearrangement factor to obtain the rearrangement result. The rearrangement result is a combination of factors such as click-through rate, conversion rate, some national policies, company strategy and operation rules. Therefore, the result can be directly displayed to users so that they can choose their favorite products.

[0066] As can be seen, this embodiment of the application obtains user requests, searches based on user requests to obtain multiple matching products; inputs multiple matching products into an initial ranking model to obtain multiple products to be re-ranked; obtains the initial ranking score of each product to be re-ranked, and obtains the ranking factor of each product to be re-ranked based on the initial ranking score; obtains a preset support factor, compares each ranking factor with the preset support factor, and selects the largest one as the re-ranking factor of each product to be re-ranked; performs a re-ranking operation on each product to be re-ranked based on each re-ranking factor, and displays the re-ranked result to the user. The above operation adopts a dynamic truncation method, calculates the re-ranking factor of each product to be re-ranked, performs a re-ranking operation on each product to be re-ranked based on the re-ranking factor, and displays the re-ranked result to the user, so as to achieve the effect of disrupting the original ranking scenario at the lowest possible cost, and ensures that the re-ranked products better meet the user's needs while ensuring click-through rate and conversion rate.

[0067] The different steps in the above method flow are described in detail below. First, step S203, namely "inputting multiple matching products into the initial ranking model to obtain multiple products to be rearranged", will be described in detail with reference to the embodiment.

[0068] Filter multiple matching products to obtain multiple filtered products; input the multiple filtered products into the coarse ranking model to obtain multiple coarse-ranked products; input the multiple coarse-ranked products into the fine ranking model to obtain multiple products to be rearranged.

[0069] The initial ranking model includes a coarse ranking model and a fine ranking model. Both the coarse ranking model and the fine ranking model are pre-set models based on the specifications, colors, prices, and target audiences of different products. The difference lies in the number of products processed. The coarse ranking model processes more products, while the fine ranking model processes relatively fewer products.

[0070] Here, after obtaining multiple matching products, a filtering operation can be performed on these products, removing those that are no longer available or have insufficient stock, resulting in multiple filtered products. Inputting these filtered products into the coarse-ranking model yields a list of products that meet the user's requirements regarding product specifications, color, price, and target audience. However, this list contains too many products. Therefore, these coarse-ranked products can be input into the fine-ranking model to obtain a more suitable and appropriately sized list of products for re-ranking. This list of products for re-ranking is then presented in a list format, satisfying not only the user's requirements for product specifications, color, price, and target audience, but also representing the products with the highest click-through and conversion rates.

[0071] The above operations, by sequentially filtering, coarsely ranking, and finely ranking multiple matching products, ensure that the resulting products to be re-ranked better meet user needs while maintaining click-through rates and conversion rates.

[0072] Next, the step S205 described above, "obtaining the sorting factor of each product to be rearranged based on each initial sorting," will be described in detail with reference to the embodiments.

[0073] In one embodiment, the maximum score in each initial ranking is obtained, and the maximum score is compared with each initial ranking score to obtain the multiple score of each product to be rearranged; a preset limit value and a preset hyperparameter value are obtained, and the sorting factor of each product to be rearranged is obtained based on the preset limit value, the preset hyperparameter value and each multiple score.

[0074] The sorting factor is adjustable within the range of [0,1] and is used to weight each item to be rearranged during the rearrangement process. The preset limit value is used to limit the sorting factor and can be set to 1. The preset hyperparameter value is a preset hyperparameter used to affect the control strength and change efficiency.

[0075] Here, after inputting multiple matching products into the initial ranking model, the model scores each matching product, resulting in each product to be reranked having a corresponding initial ranking score. The maximum score among these initial ranking scores is the initial ranking score of the first product to be reranked. This maximum score is then compared to each of the other initial ranking scores to obtain the multiple score for each product to be reranked. Let's assume the initial ranking score of the i-th product to be reranked is denoted as 'score'. i The maximum score among the items to be rearranged is denoted as maxScore, and the multiple score of the i-th item to be rearranged is denoted as multipleScore. i The expression for the multiple fraction of the i-th item to be rearranged is as follows:

[0076] multipleScore i =maxScore / score i

[0077] Similarly, the multiplier score for each item to be rearranged can be obtained through the above expression. Simultaneously, preset limit values ​​and preset hyperparameter values ​​are obtained, and the sorting factor for each item to be rearranged is derived based on these values ​​and the multiplier scores.

[0078] The above operation, by considering various factors such as preset limit values, preset hyperparameter values, and multiples, obtains the sorting factors for each product to be rearranged, thereby minimizing the impact on the initial sorting results.

[0079] In one embodiment, a first control factor for each product to be rearranged is obtained based on each multiple and a preset protrusion value; a second control factor is obtained based on a preset control strength value and a preset protrusion value; and a ranking factor for each product to be rearranged is obtained based on a preset limit value, each of the first control factor and the second control factor.

[0080] The preset hyperparameter values ​​include a preset control strength value and a preset spurt degree value. The preset control strength value, represented by β, influences the control strength and can be set to 500. The preset spurt degree value, represented by n, controls the degree of "spurt" and can be set to 1. Both the preset control strength value and the preset spurt degree value are intended to achieve control efficiency. Of course, the settings of the preset hyperparameter values ​​will vary in different scenarios and need to be determined based on the observation results of specific experiments.

[0081] Here, based on the multiples and preset protrusion values, the first control factor for each item to be rearranged is obtained. Let the first control factor for the i-th item to be rearranged be denoted as `firstFactor`. i Then the expression for the first regulatory factor of the i-th product to be rearranged is:

[0082] firstFactor i =multipleScore i n

[0083] The second control factor is obtained based on the preset control strength value and the preset surge degree value. It should be noted that the second control factor is the same for all products to be rearranged, and is calculated based on the hyperparameter values ​​preset by the staff. Assuming the second control factor is denoted as `secondFactor`, the expression for the second control factor of the products to be rearranged is:

[0084] secondFactor ＝ β n

[0085] The above expressions can be used to obtain the first and second control factors of each product to be rearranged. Then, based on the preset limit values, the first and second control factors, the sorting factors of each product to be rearranged can be obtained to ensure that the rearranged products better meet the needs of users.

[0086] In one embodiment, the first control factor is compared with the second control factor to obtain the control value of each product to be rearranged; the preset limit value is compared with each control value to obtain the ranking factor of each product to be rearranged.

[0087] Here, the first regulatory factor is compared with the second regulatory factor to obtain the regulatory value for each product to be rearranged. Let the regulatory value for the i-th product to be rearranged be denoted as `regulatoryValue`. i The expression for the adjustment value of the i-th item to be rearranged is as follows:

[0088] regulatoryValue i =firstFactor i / secondFactor

[0089] Similarly, the control values ​​for each item to be rearranged can be obtained through the above expressions. Then, the difference between the preset limit value and each control value is calculated to obtain the ranking factor for each item to be rearranged. The expression for the ranking factor of the i-th item to be rearranged is:

[0090] sortingFactor i =1-regulatoryValue i

[0091] =firstFactor i / secondFactor

[0092] =1-[multipleScore] i n / (β n )]

[0093] =1-[(maxScore / score)] i ) n / (β n )]

[0094] Similarly, the ranking factor of each item to be rearranged can be obtained through the above expression. It should be noted that the above expression is based on algorithmic derivation, i.e., the adjustment range of the ranking factor is [0,1]. Assuming x represents the multiple of each item to be rearranged, and y represents the ranking factor, then when the preset adjustment strength value β equals the multiple x, the ranking factor y equals 0; when the multiple x equals 1, the ranking factor y equals 1. And the preset protrusion value is represented by n. Then the expression can be established as:

[0095] y = a*pow(x,n) + b

[0096] a + b = 1 & a * pow(β,n) + b = 0

[0097] Therefore, we get: a = -[1 / (β) n -1)]&b=1+[1 / (β n -1)]

[0098] y = -[x n / (β n -1)]+[β n / (β n -1)]=(β n -x n ) / (β n -1)=1-[(x n -1) / (β n -1)]

[0099] Where a and b are coefficients in the formula and have no specific meaning; x represents the multiple of the goods to be rearranged; y represents the sorting factor; β represents the preset control strength value; and n represents the preset protrusion degree value. Because x n and β n Since these are all relatively large numbers, the subtraction of 1 can be omitted. Therefore, we can obtain the above expression, which is...

[0100] sortingFactor i =1-[(maxScore / score)] i ) n / (β n )]

[0101] The above operations can achieve the effect of disrupting the original sorting at the lowest possible cost, while ensuring click-through rate and conversion rate, and personalized control over the order of the rearranged products.

[0102] Finally, the following detailed description of step S209, "rearranging each item to be rearranged according to each rearrangement factor and displaying the rearranged result to the user," will be provided in conjunction with the embodiments.

[0103] In one embodiment, the initial ranking score of each item to be rearranged is multiplied by its corresponding rearrangement factor to obtain the weighted score of each item to be rearranged; the items to be rearranged are rearranged based on their initial ranking score and corresponding weighted score, and the rearranged result is displayed to the user.

[0104] Here, we assume that the weighted score of the i-th item to be rearranged is represented as weightedScore. i Then its specific expression is:

[0105] weightedScore i =Score i* rearrangementFactor i

[0106] Similarly, the weighted score of each item to be rearranged can be obtained through the above expression. Then, based on the initial score and corresponding weighted score of each item, the items to be rearranged are rearranged, and the rearranged result is displayed to the user.

[0107] The above operation reorders the products based on their initial scores and corresponding weighted scores, and displays the reordered results to the user to ensure that the reordered products better meet the user's needs.

[0108] In one embodiment, the initial ranking score of each item to be rearranged is summed with its corresponding weighted score to obtain the rearrangement score of each item; the items to be rearranged are then rearranged in descending order of their rearrangement scores, and the rearranged results are displayed to the user.

[0109] Here, we assume that the rearrangement score of the i-th item to be rearranged is represented as rearrangementScore. i Then its specific expression is:

[0110] rearrangementScore i =Score i +Score i * rearrangementFactor i

[0111] =Score i* (1+rearrangementFactor i )

[0112] Similarly, using the above expression, the reorder score of each product to be reordered can be obtained. The products to be reordered are then rearranged in reverse order, i.e., in descending order of their reorder scores, to obtain a sorting result that takes into account the reordering efficiency factor. This result is then displayed to the user, thereby solving the problem of the unreasonable "one-size-fits-all" approach in the existing technology. This achieves the effect of minimizing the impact on the initial sorting result, and while ensuring click-through rate and conversion rate, it ensures that the reordered products better meet the user's needs.

[0113] Based on the implementation methods in the above embodiments, the following will be combined with... Figure 3 A preferred method flow provided in an embodiment of this application will be described by way of example. For instance... Figure 3 As shown, the method may include the following steps:

[0114] Step S301: Obtain the user request, perform a search based on the user request, and obtain multiple matching products.

[0115] Step S302: Filter the multiple matching products to obtain multiple filtered products.

[0116] Step S303: Input multiple filtered products into the coarse-ranking model to obtain multiple coarse-ranked products.

[0117] Step S304: Input multiple coarse-sorted items into the fine-sorting model to obtain multiple items to be rearranged.

[0118] Step S305: Obtain the initial sorting score for each item to be rearranged.

[0119] Step S306: Obtain the maximum score in each initial ranking score, and perform a ratio calculation between the maximum score and each initial ranking score to obtain the multiple score for each item to be rearranged.

[0120] Step S307: Based on each multiple and the preset protrusion value, the first control factor of each product to be rearranged is obtained.

[0121] Step S308: Obtain the second control factor based on the preset control strength value and the preset protrusion degree value.

[0122] Step S309: Calculate the ratio of each first regulatory factor to the second regulatory factor to obtain the regulatory value of each commodity to be rearranged.

[0123] Step S310: Perform difference calculations between the preset limit value and each control value to obtain the sorting factor of each product to be rearranged.

[0124] Step S311: Obtain the preset support factor, compare each sorting factor with the preset support factor, and select the largest one as the sorting factor for each product to be sorted.

[0125] Step S312: Multiply the initial score of each item to be rearranged by its corresponding rearrangement factor to obtain the weighted score of each item to be rearranged.

[0126] Step S313: Add the initial score of each item to be rearranged to its corresponding weighted score to obtain the rearrangement score of each item.

[0127] Step S314: Rearrange the corresponding products to be rearranged according to the order of each rearrangement score from high to low, and display the rearranged results to the user.

[0128] It should be understood that, although Figures 2-3 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated in this application, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Furthermore, Figures 2-3 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0129] The above method embodiments can be applied to a variety of application scenarios, such as scenarios that include, but are not limited to, reordering products to be rearranged after initial sorting.

[0130] Figure 4 This is a schematic diagram of a product rearrangement device provided in an embodiment of this application. The device can be installed in... Figure 1 The server in the application environment shown is used to perform, for example... Figures 2-3 The method flow is shown below. Figure 4 As shown, the device may include: a matching module 401, an initial sorting module 403, a sorting factor acquisition module 405, a rearrangement factor acquisition module 407, and a rearrangement module 409. The main functions of each component module are as follows:

[0131] The matching module 401 is used to obtain user requests, search based on user requests, and obtain multiple matching products;

[0132] The initial sorting module 403 is used to input multiple matching products into the initial sorting model to obtain multiple products to be rearranged.

[0133] The sorting factor acquisition module 405 is used to acquire the initial sorting score of each product to be rearranged, and to obtain the sorting factor of each product to be rearranged based on the initial sorting score.

[0134] The module 407 for obtaining rearrangement factors is used to obtain preset support factors, compare each sorting factor with the preset support factors, and select the largest one as the rearrangement factor for each product to be rearranged.

[0135] The rearrangement module 409 is used to rearrange each product to be rearranged according to each rearrangement factor and display the rearranged result to the user.

[0136] In one embodiment, the initial sorting model includes a coarse sorting model and a fine sorting model, and the initial sorting module 403 is further used for:

[0137] Filter multiple matching products to obtain multiple filtered products;

[0138] Input multiple filtered products into the coarse-ranking model to obtain multiple coarse-ranked products;

[0139] Input multiple coarse-ranked items into the fine-ranking model to obtain multiple items to be rearranged.

[0140] In one embodiment, the sorting factor acquisition module 405 is further configured to:

[0141] Obtain the maximum score from each initial ranking score, and calculate the ratio of the maximum score to each initial ranking score to obtain the multiple score for each item to be rearranged.

[0142] Obtain the preset limit value and preset hyperparameter value, and based on the preset limit value, preset hyperparameter value and each multiplier, obtain the sorting factor for each product to be rearranged.

[0143] In one embodiment, the preset hyperparameter values ​​include a preset control strength value and a preset protrusion degree value. The sorting factor acquisition module 405 is further used for:

[0144] Based on the multiples and the preset protrusion values, the first control factor for each product to be rearranged is obtained;

[0145] The second control factor is obtained based on the preset control strength value and the preset protrusion degree value;

[0146] Based on the preset limit value, each first control factor, and the second control factor, the sorting factor for each product to be rearranged is obtained.

[0147] In one embodiment, the sorting factor acquisition module 405 is further configured to:

[0148] The ratios of each first regulatory factor to the second regulatory factor are calculated to obtain the regulatory values ​​for each commodity to be rearranged.

[0149] The difference between the preset limit value and each control value is calculated to obtain the sorting factor of each product to be rearranged.

[0150] In one embodiment, the rearrangement module 409 is further configured to:

[0151] The initial score of each item to be rearranged is multiplied by its corresponding rearrangement factor to obtain the weighted score of each item to be rearranged.

[0152] The products to be rearranged are rearranged based on their initial ranking score and corresponding weighted score, and the rearranged results are displayed to the user.

[0153] In one embodiment, the rearrangement module 409 is further configured to:

[0154] The initial ranking score of each item to be rearranged is summed with its corresponding weighted score to obtain the rearrangement score of each item to be rearranged.

[0155] The products to be rearranged are rearranged in descending order of their rearrangement scores, and the results are displayed to the user.

[0156] The same or similar parts among the above embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple, and the relevant parts can be referred to the description of the method embodiments.

[0157] It should be noted that the embodiments of this application may involve the use of user data. In practical applications, user-specific personal data may be used in the scheme described herein within the scope permitted by applicable laws and regulations, provided that it complies with the applicable laws and regulations of the country (e.g., explicit consent from the user, actual notification to the user, explicit authorization from the user, etc.).

[0158] According to embodiments of this application, this application also provides a computer device and a computer-readable storage medium.

[0159] like Figure 5 The diagram shown is a block diagram of a computer device according to an embodiment of this application. The term "computer device" is intended to represent various forms of digital computers or mobile devices. The digital computer may include a desktop computer, a portable computer, a workbench, a personal digital assistant, a server, a mainframe computer, and other suitable computers. The mobile device may include a tablet computer, a smartphone, a wearable device, etc.

[0160] like Figure 5 As shown, device 500 includes a computing unit 501, a ROM 502, a RAM 503, a bus 504, and an input / output (I / O) interface 505. The computing unit 501, ROM 502, and RAM 503 are interconnected via the bus 504. The input / output (I / O) interface 505 is also connected to the bus 504.

[0161] The computing unit 501 can execute various processes in the method embodiments of this application according to computer instructions stored in the read-only memory (ROM) 502 or computer instructions loaded from the storage unit 508 into the random access memory (RAM) 503. The computing unit 501 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. The computing unit 501 can include, but is not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. In some embodiments, the methods provided in the embodiments of this application can be implemented as computer software programs, which are tangibly contained in a computer-readable storage medium, such as the storage unit 508.

[0162] RAM 503 can also store various programs and data required for the operation of device 500. Part or all of the computer program can be loaded and / or installed on device 500 via ROM 502 and / or communication unit 509.

[0163] The input unit 506, output unit 507, storage unit 508, and communication unit 509 in device 500 can be connected to I / O interface 505. The input unit 506 can be, for example, a keyboard, mouse, touchscreen, or microphone; the output unit 507 can be, for example, a display, speaker, or indicator light. Device 500 can exchange information and data with other devices through the communication unit 509.

[0164] It should be noted that the device may also include other components necessary for normal operation. It may also include only the components necessary for implementing the solution of this application, without necessarily including all the components shown in the figures.

[0165] Various implementations of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SOCs), payload programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof.

[0166] The computer instructions used to implement the methods of this application may be written in any combination of one or more programming languages. These computer instructions may be provided to the computing unit 501 such that when executed by the computing unit 501, such as a processor, the computer instructions cause the execution of the steps involved in the embodiments of the methods of this application.

[0167] The computer-readable storage medium provided in this application can be a tangible medium that can contain or store computer instructions for performing the steps involved in the method embodiments of this application. The computer-readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, and other forms of storage media.

[0168] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for rearranging goods, characterized in that, The method includes: Obtain user requests, search based on user requests, and obtain multiple matching products; Input the multiple matched products into the initial ranking model to obtain multiple products to be rearranged; Obtain the initial ranking score of each of the items to be rearranged, and the maximum score among the initial ranking scores; The maximum score is compared with each of the initial scores to obtain the multiple score of each of the items to be rearranged; Obtain preset limit values ​​and preset hyperparameter values, wherein the preset hyperparameter values ​​include preset control force values ​​and preset protrusion degree values; Based on the multiplier and the preset protrusion value, the first control factor of each of the products to be rearranged is obtained; The second control factor is obtained based on the preset control strength value and the preset protrusion degree value; Based on the preset limit value, each of the first control factors and the second control factors, the sorting factor of each of the products to be rearranged is obtained; Obtain a preset support factor, compare each sorting factor with the preset support factor, and select the largest one as the sorting factor for each item to be sorted. The rearrangement operation is performed on each of the items to be rearranged according to the rearrangement factors, and the rearranged results are displayed to the user.

2. The method according to claim 1, characterized in that, The initial ranking model includes a coarse ranking model and a fine ranking model. Multiple matching products are input into the initial ranking model to obtain multiple products to be rearranged, including: The multiple matching products are filtered to obtain multiple filtered products; Inputting multiple filtered products into the coarse-ranking model yields multiple coarse-ranked products; Inputting multiple coarsely ranked products into the finely ranked model yields multiple products to be rearranged.

3. The method according to claim 1, characterized in that, The step of obtaining the sorting factor for each of the items to be rearranged based on the preset limit value, each of the first control factors, and the second control factor includes: The first regulatory factor is compared with the second regulatory factor to obtain the regulatory value of each of the commodities to be rearranged. The preset limit value and each of the control values ​​are used to perform a difference operation to obtain the sorting factor of each of the products to be rearranged.

4. The method according to any one of claims 1-3, characterized in that, The step of rearranging each of the items to be rearranged according to each of the rearrangement factors and displaying the rearranged results to the user includes: The initial score of each item to be rearranged is multiplied by the corresponding rearrangement factor to obtain the weighted score of each item to be rearranged. The products to be rearranged are rearranged based on their initial scores and corresponding weighted scores, and the rearranged results are displayed to the user.

5. The method according to claim 4, characterized in that, The step of rearranging the items to be rearranged based on their initial ranking score and corresponding weighted score, and then displaying the rearranged results to the user, includes: The initial ranking score of each item to be rearranged is summed with the corresponding weighted score to obtain the rearrangement score of each item to be rearranged. The items to be rearranged are rearranged in descending order of their respective rearrangement scores, and the rearranged results are displayed to the user.

6. A product rearrangement device, characterized in that, The device includes: The matching module is used to obtain user requests, search based on the user requests, and obtain multiple matching products; The initial ranking module is used to input multiple matching products into the initial ranking model to obtain multiple products to be rearranged. A sorting factor acquisition module is used to acquire the initial ranking score of each item to be rearranged, and the maximum score among the initial ranking scores; to perform a ratio calculation between the maximum score and each of the initial ranking scores to obtain the multiple score of each item to be rearranged; to acquire a preset limit value and a preset hyperparameter value, wherein the preset hyperparameter value includes a preset control strength value and a preset protrusion degree value; to acquire a first control factor for each item to be rearranged based on each of the multiple scores and the preset protrusion degree value; to acquire a second control factor based on the preset control strength value and the preset protrusion degree value; and to acquire a sorting factor for each item to be rearranged based on the preset limit value, each of the first control factors, and the second control factor. The module for obtaining rearrangement factors is used to obtain a preset support factor, compare each of the sorting factors with the preset support factor, and select the largest one as the rearrangement factor for each of the products to be rearranged. The rearrangement module is used to rearrange each of the items to be rearranged according to each of the rearrangement factors, and to display the rearranged results to the user.

7. A computer device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores computer instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method of any one of claims 1-5.

8. A computer-readable storage medium storing computer instructions thereon, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1 to 5.

Images