Commodity combination recommendation method and computer equipment

Abstract

The invention provides a commodity combination recommendation method and computer equipment, and the method comprises the steps: responding to a combination recall instruction, determining a target commodity corresponding to the combination recall instruction, and determining a plurality of initial commodity combinations comprising the target commodity; determining user preference features and store features corresponding to the target commodity; determining commodity combination features corresponding to each initial commodity combination; based on the user preference features, the store features and the commodity combination features, determining a recommendation score corresponding to each initial commodity combination; and according to the recommendation score, determining a target commodity combination from the plurality of initial commodity combinations. According to the invention, the recommendation effect of the commodity combination can be improved.

Description

Technical Field

[0001] The present application belongs to the field of computer technology, relates to data processing technology, and in particular, to a method for recommending a product combination and a computer device. Background Art

[0002] In the scenario of providing product purchase services for users, the purchase rate of each product combination can be predicted by combining multiple products, and the product combination with the highest purchase rate can be displayed to users.

[0003] In the related art, the historical behavior data of commodities are counted to determine the behavior data such as common purchase or common click between commodities to mine commodity combinations. However, the above method ignores the diverse needs of users and is difficult to generate recommendation data that meets the personalized needs of users, resulting in poor recommendation effect of commodity combinations. Summary of the invention

[0004] The embodiments of the present application provide a method and a computer device for recommending a product combination to solve the problem that the recommendation effect of a product combination is poor due to ignoring the diverse needs of users.

[0005] In a first aspect, an embodiment of the present application provides a method for recommending a product combination, the method comprising: in response to a combination recall instruction, determining a target product corresponding to the combination recall instruction, and determining a plurality of initial product combinations including the target product; determining user preference characteristics and store characteristics corresponding to the target product; determining product combination characteristics corresponding to each initial product combination; based on the user preference characteristics, the store characteristics, and the product combination characteristics, determining a recommendation score corresponding to each initial product combination; and determining a target product combination from the plurality of initial product combinations based on the recommendation score.

[0006] In some feasible embodiments, the method further includes: determining a preset recall source; and determining the multiple initial commodity combinations corresponding to the target commodity based on the recall source. In some feasible embodiments, the method further includes: determining the combination score corresponding to each initial product combination based on the recall source; determining the combination attribute corresponding to each initial product combination; selecting the initial product combination corresponding to the combination score whose combination attribute is a preset attribute and is greater than or equal to a preset score threshold, and determining the product combination feature corresponding to the selected initial product combination.

[0007] In some feasible embodiments, determining the user preference characteristics and store characteristics corresponding to the target product includes: determining the target user of the target product and the preferred product information corresponding to the target user; determining the user preference characteristics based on the preferred product information; determining the target store of the target product and the store information corresponding to the target store; and determining the store characteristics based on the store information.

[0008] In some feasible embodiments, determining the user preference characteristics based on the preferred product information includes: determining a product list and determining a product category of a preset level corresponding to each product in the product list based on the preferred product information to obtain a category list; vectorizing each product in the product list to obtain multiple product vectors; vectorizing each product category in the category list to obtain multiple category vectors; and determining the user preference characteristics based on the multiple product vectors and the multiple category vectors.

[0009] In some feasible embodiments, determining the store characteristics based on the store information includes: determining store attribute information and store operation information based on the store information; vectorizing the store attribute information to obtain store attribute characteristics; normalizing the store operation information to obtain store operation characteristics; determining the store characteristics based on the store attribute characteristics and the store operation characteristics. In some feasible embodiments, determining the commodity combination characteristics corresponding to each initial commodity combination includes: determining the combination dimension information and combination statistical information corresponding to each initial commodity combination; vectorizing the combination dimension information to obtain a combination dimension vector; normalizing the combination statistical information to obtain a combination statistical vector; and determining the commodity combination characteristics based on the combination dimension vector and the combination statistical vector.

[0010] In some feasible embodiments, determining the recommendation score corresponding to each initial product combination based on the user preference characteristics, the store characteristics and the product combination characteristics includes: determining the behavior information corresponding to each product in each initial product combination based on the user preference characteristics, the store characteristics and the product combination characteristics; determining the recommendation score corresponding to each initial product combination based on the behavior information. In some feasible embodiments, selecting and recommending a target product combination from the multiple initial product combinations based on the recommendation score includes: sorting the multiple initial product combinations based on the recommendation score and a preset sorting rule, and determining a preset number of target product combinations ranked first as the target product combination; and pushing recommendation information corresponding to the target product combination to the terminal device of the target user.

[0011] In a second aspect, an embodiment of the present application provides a device for recommending a product combination, the device comprising: a combination recall module, for determining, in response to a combination recall instruction, a target product corresponding to the combination recall instruction, and determining a plurality of initial product combinations including the target product; a feature determination module, for determining a user preference feature and a store feature corresponding to the target product; the feature determination module is also used to determine a product combination feature corresponding to each initial product combination; a score determination module, for determining a recommendation score corresponding to each initial product combination based on the user preference feature, the store feature and the product combination feature; a combination determination module, for determining a target product combination from the plurality of initial product combinations based on the recommendation score.

[0012] In a third aspect, an embodiment of the present application provides a computer device, comprising a controller and a memory, wherein the controller is used to implement any one of the above-mentioned methods for recommending a combination of goods when executing a computer program stored in the memory.

[0013] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a controller, the method for recommending a product combination described in any one of the above is implemented.

[0014] The embodiment of the present application provides a method for recommending a product combination, which, in response to a combination recall instruction, determines a plurality of initial product combinations of target products corresponding to the combination recall instruction; determines the user preference characteristics and store characteristics corresponding to the target products; determines the product combination characteristics corresponding to each initial product combination; determines the recommendation score corresponding to each initial product combination based on the user preference characteristics, the store characteristics, and the product combination characteristics; and selects and recommends a target product combination from the plurality of initial product combinations based on the recommendation score. The above method determines the target product combination to be recommended based on the user preference characteristics, the store characteristics, and the product combination characteristics by determining the user preference characteristics and store characteristics corresponding to the target products and determining the product combination characteristics corresponding to each initial product combination. By determining the target product combination in multiple dimensions, the effect of product combination recommendation can be improved. The above method determines the product combination based on the user preference characteristics, so that the product combination meets the diverse needs of users and improves the effect of product combination recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is an application scenario diagram of a method for recommending a product combination provided in an embodiment of the present application.

[0016] Figure 2It is a flowchart of a method for recommending a product combination provided in an embodiment of the present application.

[0017] Figure 3 It is a schematic diagram of the initial product combination determination process provided in an embodiment of the present application.

[0018] Figure 4 It is a flowchart of the method for determining the initial product combination provided in an embodiment of the present application.

[0019] Figure 5 It is a flowchart of the feature determination method provided in an embodiment of the present application.

[0020] Figure 6 It is a flowchart of the method for determining preference characteristics provided in an embodiment of the present application.

[0021] Figure 7 This is a flow chart for determining store characteristics provided in an embodiment of the present application.

[0022] Figure 8 It is a flow chart for determining product combination characteristics provided in an embodiment of the present application.

[0023] Fig. 9 It is a schematic diagram of the structure of a device for recommending a product combination provided in an embodiment of the present application.

[0024] Fig.10 This is a diagram of application equipment of the method for recommending a combination of goods provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] It should be noted that the terms "first" and "second" in the specification, claims and drawings of the present application are used to distinguish similar objects rather than to describe a specific order or sequence.

[0026] It should also be noted that the method disclosed in the embodiments of the present application or the method shown in the flowchart includes one or more steps for implementing the method. Without departing from the scope of the claims, the execution order of multiple steps can be interchangeable with each other, and some of the steps can also be deleted.

[0027] Some embodiments will be described below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments may be combined with each other.

[0028] Figure 1 : is an application scenario diagram of a method for recommending a product combination provided in an embodiment of the present application. Figure 1 As shown, the application scenario includes a computer device 10 and a user terminal 20. As a feasible embodiment, the method for recommending a product combination provided in the embodiment of the present application can be applied to the computer device 10. The computer device 10 can be as follows: Figure 1 The computer device 10 may be connected to the user terminal 20 via a network. The network is used to provide a medium for a communication link between the computer device 10 and the user terminal 20. The network may include various connection types, such as a wired communication link, a wireless communication link, etc., which is not limited in the embodiments of the present application.

[0029] Figure 1 The computer device 10, network and user terminal 20 are merely illustrative. The computer device 10 may be a physical server or a server cluster composed of multiple servers, and the user terminal 20 may be a mobile phone, a tablet, a desktop computer, a laptop computer, or other devices. As another example, the embodiment of the present application may also allow multiple user terminals 20 to access the computer device 10.

[0030] In some embodiments, the technical solution of the present application uses the operating environment of the e-commerce service platform as its application environment, and builds an e-commerce service platform in the computer device 10. The e-commerce service platform can be an e-commerce service platform that opens independent station services, such as a cross-border e-commerce service platform. An independent station refers to a new type of official website (website) established on a Software-as-a-Service (SaaS) technology platform, which has an independent domain name, private content, data, and rights and interests, independent business sovereignty and business entity responsibilities, supported by social cloud computing capabilities, and can independently and freely connect to third-party software tools, publicity and promotion media and channels. Usually, product sellers build and operate online stores on independent stations, allowing online stores to display their products on independent stations, and provide product buyers (i.e. users) with e-commerce services such as searching for products in online stores, adding shopping carts, settlement, logistics delivery, placing orders, after-sales service, and product combination recommendations.

[0031] In some embodiments, the user terminal 20 can access the online store in the computer device 10 by connecting to the computer device 10, and realize functions such as searching for products, adding shopping carts, and settlement in the online store. In some embodiments, when a user clicks on a product in the online store, the computer device 10 can determine the product combination containing the product from the online store according to the product combination recommendation method provided in the embodiment of the present application (for ease of description, the embodiment of the present application is referred to as the "target product combination"), and push the recommendation information corresponding to the target product combination to the user terminal 20.

[0032] Figure 2 is a flowchart of a method for recommending a product combination provided in an embodiment of the present application, wherein the method for recommending a product combination is applied to a computer device (for example, Figure 1 The computer device 10 in FIG. Figure 2As shown, the method for recommending a product combination may include the following steps S11-S15. According to different requirements, the order of the steps in the flowchart may be changed, and some may be omitted.

[0033] S11, in response to a combined recall instruction, determining a target commodity corresponding to the combined recall instruction, and determining a plurality of initial commodity combinations including the target commodity.

[0034] In at least one embodiment of the present application, the combination recall instruction may be an instruction for recalling a product combination of a product (referred to as the "target product" in the present application for ease of description) when a user clicks, adds to cart, or purchases a product in the product interface of an online store. After receiving the combination recall instruction, the target product corresponding to the combination recall instruction is determined, and a preset recall source is called to perform a recall process on the product combination corresponding to the target product, thereby obtaining multiple initial product combinations including the target product, each of which contains at least two products.

[0035] In some embodiments, the recall source may represent a recall strategy for a product combination. Each recall source may be able to recall an initial product combination including a target product, and each recall source may be able to determine a combination score corresponding to the initial product combination. The combination score is used to identify the probability of the initial product combination being recalled. The higher the combination score, the higher the probability of the initial product combination being recalled; the lower the combination score, the lower the probability of the initial product combination being recalled.

[0036] In some embodiments, the number of recall sources can be set according to actual needs, for example, the number of recall sources can be 1, 3, 5, etc. The embodiment of the present application takes the number of recall sources as 3 as an example for explanation. For example, the recall sources may include heat recall sources, similarity recall sources, and user behavior recall sources, wherein the heat recall source refers to a strategy for recalling a combination of goods based on the user's click, favorite, add-to-cart, purchase, and other behavioral data, the similarity recall source refers to a strategy for determining the user's favorite combination of goods based on the user's historical behavior data, and recalling other combinations of goods similar to the favorite combination of goods based on the similarity matrix corresponding to the combination of goods, and the user behavior recall source refers to dividing users into different groups based on the user's purchase history, browsing history, search habits, and other behavioral data, each group has similar purchase or browsing patterns; formulate personalized recall strategies for different user groups. The recall process of heat recall sources, similarity recall sources, and user behavior recall sources can refer to the relevant technology and will not be repeated here. The initial product combinations recalled by different recall sources may be the same or different, and the combination scores of the initial product combinations recalled by different recall sources may be the same or different, and there is no restriction here.

[0037] S12, determining the user preference characteristics and store characteristics corresponding to the target product.

[0038] In at least one embodiment of the present application, the user who clicks, adds to cart, or purchases the target product is determined (for ease of description, this application refers to the "target user") and the store corresponding to the target product is determined (for ease of description, this application refers to the "target store"). In some embodiments, the user preference feature may represent the feature obtained by data processing the target user's preferred product information, wherein the data processing method includes vectorization processing and normalization processing. The preferred product information may include, but is not limited to, the target user's product list and category list, wherein the product list may include a preferred product list, a preferred product tag list, a browsed product list, and a browsed product tag list. The products in the preferred product list are regarded as preferred products, and the products in the browsed product list are regarded as browsed products. The category list includes a preset level of product category corresponding to each preferred product in the preferred product list and a preset level of product category corresponding to each browsed product in the browsed product list.

[0039] In some embodiments, the preferred product list represents a set of products preferred by the target user. The preferred products can be determined by the target user's behavioral data such as clicks, add-to-carts, and purchases of products on the e-commerce service platform to obtain a preferred product list. Exemplarily, the number of clicks, add-to-carts, and purchases of each product by the target user on the e-commerce service platform is determined; based on the number of clicks, add-to-carts, and purchases of each product, the set of products preferred by the target user is determined. For example, the sum of the number of clicks, add-to-carts, and purchases corresponding to each product is determined. The larger the sum, the higher the target user's preference for the product, and the smaller the sum; the lower the target user's preference for the product.

[0040] In some embodiments, the preferred commodity list includes multiple preferred commodities, and each preferred commodity has a corresponding commodity category, and the commodity category can represent the type or type of commodity. Each preferred commodity has a preset level of commodity categories, and the preset level can be set according to actual needs. For example, the preset level can include a parent category and a first-level category. The set of commodity categories corresponding to the preferred commodities is used as the category list of the preferred commodities, the set of parent categories corresponding to the preferred commodities is used as the parent category list of the preferred commodities, and the set of first-level categories corresponding to the preferred commodities is used as the first-level category list of the preferred commodities. Exemplarily, when the category of the preferred commodity is short-sleeved, the corresponding parent category is tops, and the corresponding first-level category is clothing.

[0041] In some embodiments, the browsed product list represents a set of products browsed by the target user. The browsed products can be determined by the target user's browsing behavior on the products in the e-commerce service platform to obtain the browsed product list. The browsed product list contains multiple browsed products, and each browsed product has a corresponding preset level of product category. The preset level can be set according to actual needs. For example, the preset level can include a parent category and a first-level category. The set of product categories corresponding to the browsed products is used as the category list of the browsed products, the set of parent categories corresponding to the browsed products is used as the parent category list of the browsed products, and the set of first-level categories corresponding to the browsed products is used as the first-level category list of the browsed products. Exemplarily, when the category of the browsed product is shorts, the corresponding parent category is pants, and the corresponding first-level category is clothing.

[0042] In some embodiments, the product label may be a label indicating the product name, weight, volume, purpose, etc. The product label of the target user's preferred product is determined, and the product labels of the plurality of preferred products are used as a preferred product label list. The product label of the target user's browsed product is determined, and the product labels of the plurality of browsed products are used as a browsed product label list.

[0043] In some embodiments, the store characteristics may represent the characteristics obtained by data processing the store information of the target store, and the store information may include the store attribute information and store operation information corresponding to the target store, wherein the data processing method includes vectorization processing and normalization processing. The store attribute information may include but is not limited to the store code, the store's operating area code, the store's operating country code, the store level, the price of goods in the store, the store's main first-level category, the store's main second-level category, and the store's main detailed category. The store operation information may include but is not limited to the store order volume, the store transaction amount, and the number of store users.

[0044] S13, determining the commodity combination characteristics corresponding to each initial commodity combination.

[0045] In at least one embodiment of the present application, the commodity combination feature can be obtained by performing data processing on the commodity combination information of the initial commodity combination, and the data processing method includes vectorization processing and normalization processing. The commodity combination information may include combination dimension information and combination statistical information of the initial commodity combination. In some embodiments, the combination dimension information may include a commodity code group corresponding to the initial commodity combination, a commodity primary category group corresponding to the initial commodity combination, a commodity sub-category group corresponding to the initial commodity combination, a commodity parent category group corresponding to the initial commodity combination, and a commodity price group corresponding to the initial commodity combination. Among them, the commodity code group may represent a set of multiple commodity codes in the initial commodity combination, the commodity primary category group may represent a set of multiple commodities corresponding to the primary category in the initial commodity combination, the commodity sub-category group may represent a set of multiple commodities corresponding to the sub-category in the initial commodity combination, the commodity parent category group may represent a set of commodities corresponding to the parent category in the initial commodity combination, and the commodity price group may represent a set of commodity prices in the initial commodity combination. In some embodiments, the combined statistical information may include first statistical information obtained based on the behavioral data at the commodity level in the initial commodity combination and second statistical information obtained based on the behavioral data at the initial commodity combination level, wherein the first statistical information at the commodity level may include the first exposure and the first operation volume corresponding to each commodity in the initial commodity combination within a preset time period, and the first operation volume may include the first click volume and the first click rate, the first additional purchase volume and the first additional purchase rate, and the first purchase volume and the first purchase rate. The preset time period may be set according to actual needs, for example, the preset time period may be multiple, including 1 day, 7 days and 14 days respectively. The first exposure volume may indicate the number of times the commodity is displayed. The first click rate may be determined based on the first click volume and the number of times the commodity is displayed, the first additional purchase rate may be determined based on the first additional purchase volume and the number of times the commodity is displayed, and the first purchase rate may be determined based on the first purchase volume and the number of times the commodity is displayed. The second statistical information at the initial commodity combination level may include the second exposure and the second operation volume corresponding to the initial commodity combination within the preset time period, and the second operation volume may include the second click volume and the second click rate, the second additional purchase volume and the second additional purchase rate, and the second purchase volume and the second purchase rate. The second exposure volume may indicate the number of times the initial commodity combination is displayed. The second click rate can be determined based on the second click volume and the number of times the product is displayed, the second add-to-cart rate can be determined based on the second add-to-cart volume and the number of times the product is displayed, and the second purchase rate can be determined based on the second purchase volume and the number of times the product is displayed.

[0046] S14, determining a recommendation score corresponding to each initial product combination based on the user preference characteristics, the store characteristics and the product combination characteristics.

[0047] In at least one embodiment of the present application, the determination of the recommendation score corresponding to each initial product combination based on the user preference feature, the store feature and the product combination feature may include: determining the behavior information corresponding to each product in each initial product combination based on the user preference feature, the store feature and the product combination feature; and determining the recommendation score corresponding to each initial product combination according to the behavior information. Wherein, the behavior information is used to indicate whether the product is clicked. In other embodiments, the behavior information may also be used to indicate whether the product is added to a purchase or purchased. The present application embodiment uses the behavior information to indicate whether the product is clicked for explanation. Whether the product is clicked may be determined by a preset value, and the preset value may be 0 or 1, wherein 0 indicates that the product is not clicked and 1 indicates that the product is clicked. The recommendation score corresponding to the initial product combination is determined by determining the number of clicked products in the initial product combination. If the number of clicked products in the initial product combination is greater, the recommendation score corresponding to the initial product combination is higher; if the number of clicked products in the initial product combination is smaller, the recommendation score corresponding to the initial product combination is lower.

[0048] In some embodiments, a score determination model is pre-trained, and the score determination model is used to determine the recommendation score corresponding to the initial commodity combination based on the user preference characteristics, store characteristics and commodity combination characteristics of the initial commodity combination. The input data of the score determination model can be the initial commodity combination code, the user code corresponding to the initial commodity combination, the store code and the target commodity code, and the output data is the behavior information corresponding to each commodity in the initial commodity combination and the recommendation score corresponding to the initial commodity combination. Among them, the initial commodity combination code is used to identify the initial commodity combination, the user code is used to identify the user's identity information, the store code is used to identify the store information, and the target commodity code is used to identify the target commodity. In some embodiments, the score determination model may include but is not limited to convolutional neural network models such as LeNet-5, AlexNet, VGGNet, ResNet, Inception, or supervised learning models such as support vector machines, and this application is not limited to this. In some embodiments, before training the score determination model, the training data can be determined by burying points and the like, and the training data may include the recalled commodity combination code, the user code corresponding to the commodity combination, the store code, the target commodity code, and the behavior information corresponding to the commodity combination. Among them, the behavior information corresponding to the commodity combination may include behavior labels such as clicks, add-on purchases, and purchases of commodities in the commodity combination. The product combination code, the user code corresponding to the product combination, the store code, and the target product code are used as input data, and the behavior information and recommendation score corresponding to the product combination are used as output data for model training to obtain a score determination model. Among them, the recommendation score corresponding to the product combination can be obtained by analyzing the behavior information of the product combination through preset algorithm rules. Based on the score determination model, the present application can determine the user preference characteristics, store characteristics, and product combination characteristics corresponding to the initial product combination, and predict the behavior information of the products in the initial product combination based on the user preference characteristics, store characteristics, and product combination characteristics, and determine the recommendation score corresponding to the initial product combination.

[0049] S15: determining a target commodity combination from the plurality of initial commodity combinations according to the recommendation scores.

[0050] In at least one embodiment of the present application, multiple initial product combinations are ranked according to the recommendation scores, the initial product combinations ranked first are selected as the target product combinations, and the target product combinations are recommended to the user. The number of target product combinations can be one or more, and the present application takes the number of target product combinations as an example for explanation.

[0051] An embodiment of the present application provides a method for recommending product combinations, by determining user preference characteristics and store characteristics corresponding to target products and determining product combination characteristics corresponding to each initial product combination, and determining a target product combination to be recommended based on the user preference characteristics, the store characteristics and the product combination characteristics. By determining the target product combination in multiple dimensions, the effect of product combination recommendation can be improved. The above method determines the product combination based on the user preference characteristics, so that the product combination meets the diverse needs of users and improves the effect of product combination recommendation.

[0052] In at least one embodiment of the present application, a recall source is pre-set, and multiple initial product combinations including the target product are recalled through the recall source. In some embodiments, the method further includes: determining a preset recall source; and determining the multiple initial product combinations corresponding to the target product based on the recall source. In some embodiments, the recall source may represent the recall strategy of a product combination, and the number of the recall sources may be 1, 3, 5, etc. The embodiment of the present application takes the number of the recall sources as 3 as an example for explanation. Figure 3 is a schematic diagram of the initial product combination determination process provided by the embodiment of the present application, such as Figure 3 As shown, the recall source may include a heat recall source, a similarity recall source, and a user behavior recall source. The initial product combinations including the target product recalled by the heat recall source are initial product combination A1 and initial product combination A2. The initial product combinations including the target product recalled by the similarity recall source are initial product combination B1. The initial product combinations including the target product recalled by the user behavior recall source are initial product combination C1, initial product combination C2, and initial product combination C3. Based on this, the number of initial product combinations corresponding to the target product is 6, namely initial product combination A1, initial product combination A2, initial product combination B1, initial product combination C1, initial product combination C2, and initial product combination C3.

[0053] In at least one embodiment of the present application, the recall source may recall multiple initial product combinations and determine a combination score corresponding to each initial product combination. Based on the combination score of each initial product combination and the combination attributes of the initial product combination, initial product combinations with low or invalid combination scores may be filtered out from the multiple initial product combinations. Figure 4 1 is a flow chart of a method for determining an initial product combination provided in an embodiment of the present application, and the method for determining an initial product combination is applied to a computer device. Figure 4 As shown, the following steps are included: S21: Determine a combination score corresponding to each initial product combination based on the recall source.

[0054] In at least one embodiment of the present application, the initial product combinations recalled by different recall sources may be the same or different, and the combination scores of the initial product combinations recalled by different recall sources may be the same or different. When the combination scores of the initial product combinations recalled by different recall sources are different, the combination score corresponding to the initial product combination may be determined according to a preset selection rule. The preset selection rule may be set according to actual needs. For example, the preset selection rule may be to select a higher combination score as the combination score of the initial product combination.

[0055] S22, determining the combination attribute corresponding to each initial product combination.

[0056] In at least one embodiment of the present application, the combination attribute may include a listing attribute and a delisting attribute. The listing attribute may indicate that all commodities in the initial commodity combination are listing commodities, and the delisting attribute may indicate that there are delisted commodities in the initial commodity combination. In some embodiments, determining the combination attribute corresponding to each initial commodity combination includes: if all commodities in the initial commodity combination are listing commodities, determining the combination attribute to be a listing attribute; if there are delisted commodities in the initial commodity combination, determining the combination attribute to be a delisting attribute. By determining the combination attribute corresponding to each initial commodity combination, the embodiments of the present application can filter invalid initial commodity combinations based on the combination attributes, thereby improving the accuracy of determining the initial commodity combinations.

[0057] S23, selecting an initial commodity combination corresponding to a combination score whose combination attribute is a preset attribute and is greater than or equal to a preset score threshold, and determining the commodity combination feature corresponding to the selected initial commodity combination.

[0058] In at least one embodiment of the present application, the preset score threshold can be set according to actual needs, and an initial product combination corresponding to a combination score whose combination attribute is a preset attribute and is greater than or equal to the preset score threshold is selected from multiple initial product combinations, wherein the preset attribute can be a delisted attribute, and initial product combinations with low or invalid combination scores (for example, the combination attribute is a delisted attribute) are filtered out, thereby improving the accuracy of determining the initial product combination.

[0059] In the product combination recommendation method provided in the embodiment of the present application, based on the combination score and combination attributes of the initial product combination, the initial product combination with a low combination score or invalid is filtered out, so as to improve the recommendation effect of the product combination.

[0060] In at least one embodiment of the present application, user preference characteristics and store characteristics corresponding to the target product are determined, and a product combination to be recommended is determined by combining the user preference characteristics and the store characteristics. Figure 5 is a flow chart of a feature determination method provided in an embodiment of the present application, and the feature determination method is applied to a computer device. Figure 5As shown, the following steps are included: S31, determining the target user of the target product and the preferred product information corresponding to the target user.

[0061] In at least one embodiment of the present application, the target user may be a user who clicks, adds to cart, or purchases the target product. The correspondence between the user code and the preferred product information is preset. By determining the user code corresponding to the target user and querying the correspondence based on the determined user code, the preferred product information corresponding to the target user can be obtained. The target user's preferred product information may include, but is not limited to, a product list and a category list of the target user, wherein the product list may include a preferred product list, a preferred product tag list, a browsed product list, and a browsed product tag list. The products in the preferred product list are taken as preferred products, and the products in the browsed product list are taken as browsed products. The category list includes a preset level of product categories corresponding to each preferred product in the preferred product list and a preset level of product categories corresponding to each browsed product in the browsed product list.

[0062] S32: Determine the user preference characteristics based on the preferred product information.

[0063] In at least one embodiment of the present application, user preference features can be obtained by performing data processing on the preferred product information, and the data processing may include vectorization processing and normalization processing. Among them, vectorization processing is used to convert unstructured data (such as text) into numerical vectors, so that the semantics of the data can be understood and analyzed later. Normalization processing is used to convert data of different dimensions or ranges into a unified scale for subsequent analysis and processing. Vectorization processing and normalization processing can refer to related technologies and will not be described in detail here.

[0064] In some embodiments, determining the user preference characteristics based on the preferred product information includes: determining a data type corresponding to the preferred product information; if the data type is a numerical type, normalizing the preferred product information; if the data type is a non-numerical type, vectorizing the preferred product information.

[0065] In some embodiments, when normalizing the preferred product information, the method further includes: if the preferred product information is in a list form, normalizing each data in the list to obtain preferred product features.

[0066] In some embodiments, when the preferred product information is vectorized, the method further includes: if the preferred product information is in a list form, vectorizing each data in the list, and averaging the vectorized data to obtain the preferred product features.

[0067] Continuing from the above embodiment, the data type of the preferred product information is a non-numeric type. Taking the preferred product list in the preferred product information as an example for explanation, the preferred product list contains multiple preferred products (for example, the preferred product list contains information such as product names corresponding to multiple preferred products). Considering that the data type of the preferred product list is a non-numeric type, and the preferred product list is in list form, it is necessary to vectorize the product name of each preferred product in the preferred product list, and calculate the average of the vectorized data to obtain the product features corresponding to the preferred product list. Similarly, other lists in the preferred product information are vectorized to obtain other product features, and the multiple product features obtained are combined to obtain the preferred product features.

[0068] S33, determining the target store of the target product and the store information corresponding to the target store.

[0069] In at least one embodiment of the present application, the target store may represent an online store where the target product is located. The target store has corresponding store information. The correspondence between the store code and the store information is pre-set. By determining the store code corresponding to the target store and querying the correspondence based on the determined store code, the store information corresponding to the target store can be obtained. The store information includes store attribute information and store operation information, wherein the store attribute information may include but is not limited to the store code, the store's operating area code, the store's operating country code, the store grade, the price of goods in the store, the store's main first-level category, the store's main second-level category, and the store's main detailed category. The store operation information may include but is not limited to the store order volume, the store transaction amount, and the number of store users.

[0070] S34: Determine the store characteristics based on the store information.

[0071] In at least one embodiment of the present application, the store characteristics can be obtained by performing data processing on the store information, and the data processing may include vectorization processing and normalization processing. In some embodiments, determining the store characteristics based on the store information includes: determining the data type corresponding to the store information; if the data type is a numerical type, normalizing the store information; if the data type is a non-numerical type, vectorizing the store information.

[0072] In some embodiments, when normalizing the store information, the method further includes: if the store information is in a list form, normalizing each data in the list to obtain store features.

[0073] In some embodiments, when the store information is vectorized, the method further includes: if the store information is in a list form, each data in the list is vectorized, and the vectorized data is averaged to obtain store features.

[0074] Following the above embodiment, the store attribute information includes the store code, the store's business area code, the store's business country code, the store grade, the store's commodity price, the store's main first-level category, the store's main second-level category, and the store's main detailed category. In this way, except that the data type of the store's commodity price is a numerical type, the data type of the remaining store attribute information is a non-numerical type, wherein the store code, the store's business area code, the store's business country code, the store grade, and the store's commodity price are in a non-list form, and the store's commodity price can be normalized, and the remaining store attribute information can be vectorized to obtain store characteristics; the store's main first-level category, the store's main second-level category, and the store's main detailed category are in a list form, and each data in the list can be vectorized, and the vectorized data can be averaged to obtain store characteristics. Store business information includes store order volume, store transaction volume, and store user number. In this way, the data type of the store business information is a numerical type, and the store business information is in a non-list form, and the above data can be normalized to obtain store characteristics.

[0075] In at least one embodiment of the present application, user preference characteristics can be obtained by performing data processing on preferred product information. Figure 6 is a flow chart of a method for determining a preference feature provided in an embodiment of the present application, and the method for determining a preference feature is applied to a computer device. Figure 6 As shown, the following steps are included: S41, based on the preferred product information, determine a product list and determine a product category of a preset level corresponding to each product in the product list to obtain a category list.

[0076] In at least one embodiment of the present application, the product list may include a preferred product list, a preferred product tag list, a browsed product list, and a browsed product tag list. The products in the preferred product list are taken as preferred products, and the products in the browsed product list are taken as browsed products. The category list includes a product category of a preset level corresponding to each preferred product in the preferred product list and a product category of a preset level corresponding to each browsed product in the browsed product list.

[0077] S42, performing vectorization processing on each product in the product list to obtain multiple product vectors.

[0078] In at least one embodiment of the present application, considering that the data type of the product list is a non-numeric type, the data in the product list may be vectorized to obtain a plurality of product vectors.

[0079] S43, performing vectorization processing on each commodity category in the category list to obtain multiple category vectors.

[0080] In at least one embodiment of the present application, considering that the data type of the category list is a non-numeric type, the data in the category list can be vectorized to obtain multiple category vectors.

[0081] S44: Determine the user preference feature based on the multiple product vectors and the multiple category vectors.

[0082] In at least one embodiment of the present application, multiple product vectors and multiple category vectors are combined according to a preset data format to obtain user preference characteristics, wherein the preset data format can be set according to actual needs and is not limited here.

[0083] In the product combination recommendation method provided in the embodiment of the present application, the product list and category list in the preferred product information are vectorized to obtain product vectors and category vectors, and the product vectors and category vectors are used as user preference features. Product combinations are recommended in combination with the user preference features, so that the product combinations meet the diverse needs of users and improve the effect of product combination recommendations.

[0084] In at least one embodiment of the present application, the store information may include store attribute information and store operation information. By processing the store attribute information and the store operation information, store characteristics can be obtained. Figure 7 The flow chart of determining store characteristics provided in the embodiment of the present application is described as follows: Figure 7 As shown, determining the store characteristics based on the store information includes: determining store attribute information and store operation information based on the store information; vectorizing the store attribute information to obtain store attribute characteristics; normalizing the store operation information to obtain store operation characteristics; determining the store characteristics based on the store attribute characteristics and the store operation characteristics.

[0085] In some embodiments, the store attribute information may include, but is not limited to, the store code, the store's operating area code, the store's operating country code, the store grade, the price of goods in the store, the store's main first-level category, the store's main second-level category, and the store's main detailed category. Considering that the data type of the price of goods in the store is a numerical type, the data type of the remaining store attribute information is non-numerical. In this way, the prices of goods in the store can be normalized, and the remaining store attribute information can be vectorized to obtain store attribute features. Considering that the data type of the store's operating information is a numerical type, in this way, the store's operating information can be normalized to obtain store operating characteristics. Combining the store attribute characteristics with the store operating characteristics according to a preset data format can obtain store characteristics. Among them, the preset data format can be set according to actual needs and is not limited here.

[0086] In the product combination recommendation method provided in the embodiment of the present application, by vectorizing the store attribute information in the store information and normalizing the store operation information, the store characteristics can be obtained, and the product combination recommendation can be performed in combination with the store characteristics, which can improve the recommendation effect of the product combination. In at least one embodiment of the present application, each initial product combination has corresponding product combination information, and the product combination information includes combination dimension information and combination statistical information. By processing the combination dimension information and the combination statistical information, the product combination characteristics can be obtained. Figure 8 The flowchart for determining the characteristics of a product combination provided in the embodiment of the present application is described as follows: Figure 8 As shown, the determining of the commodity combination characteristics corresponding to each initial commodity combination includes: determining the combination dimension information and combination statistical information corresponding to each initial commodity combination; vectorizing the combination dimension information to obtain the combination dimension characteristics; normalizing the combination statistical information to obtain the combination statistical characteristics; and determining the commodity combination characteristics based on the combination dimension characteristics and the combination statistical characteristics.

[0087] In some embodiments, a correspondence between a product combination code and product combination information is preset, and the product combination information corresponding to the initial product combination can be obtained by determining the product combination code corresponding to the initial product combination and querying the correspondence based on the determined product combination code.

[0088] In some embodiments, the combination dimension information may include a commodity code group corresponding to the initial commodity combination, a commodity primary category group corresponding to the initial commodity combination, a commodity subcategory group corresponding to the initial commodity combination, a commodity parent category group corresponding to the initial commodity combination, and a commodity price group corresponding to the initial commodity combination. Considering that except for the commodity price group, the data type is numeric, and the data type of the remaining combination dimension information is non-numeric, the commodity price group can be normalized, and the combination dimension information can be vectorized to obtain the combination dimension feature.

[0089] In some embodiments, the combined statistical information may include first statistical information obtained based on behavioral data at the commodity level within the initial commodity combination and second statistical information obtained based on behavioral data at the initial commodity combination level. The first statistical information at the commodity level may include a first exposure amount and a first operation amount corresponding to each commodity in the initial commodity combination within a preset time period, and the second statistical information at the initial commodity combination level may include a second exposure amount and a second operation amount corresponding to the initial commodity combination within a preset time period. Considering that the data type of the combined statistical information is numerical, the combined statistical information may be normalized to obtain combined statistical features. Combining the combined dimensional features with the combined statistical features according to a preset data format can obtain commodity combination features. Among them, the preset data format can be set according to actual needs and is not limited here.

[0090] In the product combination recommendation method provided in the embodiment of the present application, by determining the combination dimension information and combination statistical information corresponding to the initial product combination, vectorizing the combination dimension information to obtain the combination dimension feature, and normalizing the combination statistical information to obtain the combination statistical feature, the product combination is recommended in combination with the product combination feature, which can improve the recommendation effect of the product combination.

[0091] In at least one embodiment of the present application, multiple initial commodity combinations are sorted according to the recommendation score, the initial commodity combination ranked first is selected as the target commodity combination, and the target commodity combination is recommended to the user. In some embodiments, the target commodity combination is selected and recommended from the multiple initial commodity combinations, including: sorting the multiple initial commodity combinations based on the recommendation score and the preset sorting rule, determining the target commodity combination of the preset number ranked first as the target commodity combination; pushing the recommendation information corresponding to the target commodity combination to the terminal device of the target user. Among them, the preset number can be set according to actual needs, for example, the preset number can be 1, 2, 3, etc. The recommended information may include but is not limited to the commodity price of each commodity in the target commodity combination, the total price of the commodities corresponding to the target commodity combination, and the preferential price. The recommended information can be set according to actual needs and is not limited here. In the recommendation method of the commodity combination provided in the embodiment of the present application, multiple initial commodity combinations containing the target commodity are first determined by the recall source, and then the initial commodity combination ranked first in the recommendation score is selected from the multiple initial commodity combinations as the target commodity combination in combination with the user preference characteristics, the store characteristics and the commodity combination characteristics, so that multiple commodity combinations can be recommended and the recommendation effect of the commodity combination can be improved.

[0092] See also Fig. 9 , Fig. 9 1 is a schematic diagram of the structure of the device for recommending a product combination provided in an embodiment of the present application. In some embodiments, the device for recommending a product combination 100 may include a plurality of functional modules composed of computer program segments. The computer program of each program segment in the device for recommending a product combination 100 may be stored in the memory of the computer device 10 and executed by at least one controller to execute (see Figure 2 Description) The function of recommending product combinations.

[0093] In this embodiment, the device 100 for recommending a product combination can be divided into a plurality of functional modules according to the functions it performs. The functional modules may include: a combination recall module 101, a feature determination module 102, a score determination module 103, and a combination determination module 104. The module referred to in this application refers to a series of computer program segments that can be executed by at least one controller and can complete fixed functions, which are stored in a memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0094] The combination recall module 101 may be configured to respond to a combination recall instruction, determine a target commodity corresponding to the combination recall instruction, and determine a plurality of initial commodity combinations including the target commodity.

[0095] The feature determination module 102 may be used to determine the user preference features and store features corresponding to the target product.

[0096] The feature determination module 102 is further configured to determine a commodity combination feature corresponding to each initial commodity combination.

[0097] The score determination module 103 is used to determine the recommendation score corresponding to each initial product combination based on the user preference characteristics, the store characteristics and the product combination characteristics.

[0098] The combination determination module 104 is used to determine a target commodity combination from the multiple initial commodity combinations according to the recommendation scores.

[0099] In some embodiments, the combination recall module 101 further includes: a recall source determination submodule, used to determine a preset recall source; and a product combination determination submodule, used to determine the multiple initial product combinations corresponding to the target product based on the recall source.

[0100] In some embodiments, the combination recall module 101 also includes: a combination score determination submodule, which is used to determine the combination score corresponding to each initial product combination based on the recall source; a combination attribute determination submodule, which is used to determine the combination attribute corresponding to each initial product combination; a combination selection submodule, which is used to select the initial product combination corresponding to the combination score whose combination attribute is a preset attribute and is greater than or equal to a preset score threshold, and determine the product combination characteristics corresponding to the selected initial product combination.

[0101] In some embodiments, the feature determination module 102 also includes: a preference information determination submodule, used to determine the target user of the target product and the preferred product information corresponding to the target user; a preference feature determination submodule, used to determine the user preference feature based on the preference product information; a store information determination submodule, used to determine the target store of the target product and the store information corresponding to the target store; and a store feature determination submodule, used to determine the store feature based on the store information.

[0102] In some embodiments, the feature determination module 102 also includes: a list determination submodule, which is used to determine a product list and determine the product category of a preset level corresponding to each product in the product list based on the preferred product information to obtain a category list; a product vector determination submodule, which is used to vectorize each product in the product list to obtain multiple product vectors; a category vector determination submodule, which is used to vectorize each product category in the category list to obtain multiple category vectors; and a preference feature determination submodule, which is used to determine the user preference feature based on the multiple product vectors and the multiple category vectors.

[0103] In some embodiments, the feature determination module 102 also includes: a store information parsing submodule, which is used to determine store attribute information and store operation information based on the store information; a store attribute determination submodule, which is used to vectorize the store attribute information to obtain store attribute features; an operation information determination submodule, which is used to normalize the store operation information to obtain store operation features; and a store feature determination submodule, which is used to determine the store features based on the store attribute features and the store operation features.

[0104] In some embodiments, the feature determination module 102 also includes: an information determination submodule, used to determine the combination dimension information and combination statistical information corresponding to each initial product combination; a dimension feature determination submodule, used to vectorize the combination dimension information to obtain the combination dimension feature; a statistical feature determination submodule, used to normalize the combination statistical information to obtain the combination statistical feature; and a combination feature determination submodule, used to determine the product combination feature based on the combination dimension feature and the combination statistical feature.

[0105] In some embodiments, the score determination module 103 also includes: a behavior information determination submodule, which is used to determine the behavior information corresponding to each product in each initial product combination based on the user preference characteristics, the store characteristics and the product combination characteristics; and a recommendation score determination submodule, which is used to determine the recommendation score corresponding to each initial product combination based on the behavior information.

[0106] In some embodiments, the combination determination module 104 also includes: a sorting processing submodule, which is used to sort the multiple initial product combinations based on the recommendation scores and preset sorting rules, and determine a preset number of target product combinations that are ranked first as the target product combination; a recommendation information pushing submodule, which is used to push the recommendation information corresponding to the target product combination to the terminal device of the target user.

[0107] It can be understood that the device for recommending product combinations 100 and the method for recommending product combinations in the above-mentioned embodiment belong to the same inventive concept, and the specific implementation method of each module of the device for recommending product combinations 100 corresponds to each step of the method for recommending product combinations in the above-mentioned embodiment, which will not be elaborated in this application.

[0108] The module division described above is a logical function division, and there may be other division methods in actual implementation. In addition, the functional modules in each embodiment of the present application may be integrated in the same processing unit, or each module may exist physically separately, or two or more modules may be integrated in the same unit. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0109] Fig.10 is an application device diagram of the method for recommending a product combination provided in an embodiment of the present application, such as Fig.10 As shown, the computer device 10 includes a memory 11, at least one controller 12 and at least one communication bus 13. The controller 12 is used to implement a method for recommending a combination of goods when executing a computer program stored in the memory 11. The at least one communication bus 13 is configured to implement connection and communication between the memory 11 and at least one controller 12.

[0110] Fig.10 The structure of the computer device shown does not constitute a limitation on the embodiments of the present application. The computer device 10 may also include more or less other hardware or software than shown in the figure, or a different arrangement of components.

[0111] In some embodiments of the present application, the computer device 10 can also be connected to a client device, which includes but is not limited to any electronic product that can interact with a user through a keyboard, mouse, remote control, touchpad or voice control device, such as a personal computer, a tablet computer, a smart phone, a digital camera, etc.

[0112] It should be noted that the computer device 10 is only an example, and other existing or future electronic products that are suitable for the present application should also be included in the protection scope of the present application and included here by reference.

[0113] In some embodiments, the computer device 10 may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0114] In some embodiments, a computer program is stored in the memory 11, and when the computer program is executed by at least one controller 12, all or part of the steps in the method for recommending a product combination are implemented. The memory 11 includes a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0115] In some embodiments, the computer-readable storage medium may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function, etc.; the data storage area may store data created according to the use of the computer device 10, etc.

[0116] In some embodiments, at least one controller 12 is the control core (ControlUnit) of the computer device 10, and uses various interfaces and lines to connect various components of the entire computer device 10, and executes various functions and processes data of the computer device 10 by running or executing programs or modules stored in the memory 11, and calling data stored in the memory 11. For example, when at least one controller 12 executes the computer program stored in the memory, it implements all or part of the steps of the method for recommending a combination of goods in the embodiment of the present application; or implements all or part of the functions of the device for recommending a combination of goods. At least one controller 12 can be composed of an integrated circuit, for example, it can be composed of a single packaged integrated circuit, or it can be composed of multiple integrated circuits with the same or different functions, including one or more central controllers (Central Processing unit, CPU), microcontrollers, digital processing chips, graphic controllers, and combinations of various control chips.

[0117] The above-mentioned integrated unit implemented in the form of a software function module can be stored in a computer-readable storage medium. The above-mentioned software function module is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a computer device, or a network device, etc.) or a controller (processor) to execute parts of the methods of various embodiments of the present application.

[0118] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation.

[0119] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0120] In addition, each functional module in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0121] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the spirit or basic features of the present application. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present application is limited by the attached claims rather than the above description, so it is intended to include all changes that fall within the meaning and scope of the equivalent elements of the claims in the present application. Any figure mark in the claims should not be regarded as limiting the claims involved. In addition, it is obvious that the word "including" does not exclude other units or, and the singular does not exclude the plural. Multiple units or devices stated in the specification can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical solution of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present application.

Claims

1. A method for recommending a product combination, characterized in that: The method comprises: In response to the combined recall instruction, determining a target commodity corresponding to the combined recall instruction, and determining a plurality of initial commodity combinations including the target commodity; Determine user preference characteristics and store characteristics corresponding to the target product; Determine the product combination characteristics corresponding to each initial product combination; Determining a recommendation score corresponding to each initial product combination based on the user preference characteristics, the store characteristics, and the product combination characteristics; A target commodity combination is determined from the plurality of initial commodity combinations according to the recommendation scores.

2. The method according to claim 1, characterized in that The method further comprises: Identify the intended recall sources; Based on the recall source, the multiple initial commodity combinations corresponding to the target commodity are determined.

3. The method according to claim 2, characterized in that The method further comprises: Based on the recall source, determining a combination score corresponding to each initial product combination; Determine the combination attribute corresponding to each initial product combination; An initial commodity combination corresponding to a combination score whose combination attribute is a preset attribute and is greater than or equal to a preset score threshold is selected, and the commodity combination feature corresponding to the selected initial commodity combination is determined.

4. The method according to claim 1, characterized in that The determining of the user preference characteristics and store characteristics corresponding to the target product includes: Determine the target user of the target product and the preferred product information corresponding to the target user; Determining the user preference characteristics based on the preferred product information; Determine a target store for the target product and store information corresponding to the target store; Based on the store information, the store characteristics are determined.

5. The method according to claim 4, characterized in that The determining the user preference feature based on the preferred product information includes: Based on the preferred product information, determine a product list and determine a product category of a preset level corresponding to each product in the product list to obtain a category list; Performing vectorization processing on each product in the product list to obtain multiple product vectors; Performing vectorization processing on each commodity category in the category list to obtain multiple category vectors; The user preference feature is determined based on the multiple product vectors and the multiple category vectors.

6. The method according to claim 4, characterized in that The determining the store characteristics based on the store information includes: Based on the store information, determine store attribute information and store operation information; Vectorizing the store attribute information to obtain store attribute features; Normalizing the store operation information to obtain store operation characteristics; The store characteristics are determined based on the store attribute characteristics and the store operation characteristics.

7. The method according to claim 1, characterized in that The determining of the commodity combination characteristics corresponding to each initial commodity combination includes: Determine the combination dimension information and combination statistical information corresponding to each initial product combination; Vectorizing the combined dimension information to obtain combined dimension features; Normalizing the combined statistical information to obtain combined statistical features; The commodity combination characteristics are determined based on the combination dimensional characteristics and the combination statistical characteristics.

8. The method according to claim 1, characterized in that The determining, based on the user preference feature, the store feature, and the product combination feature, a recommendation score corresponding to each initial product combination includes: Determining behavior information corresponding to each commodity in each initial commodity combination based on the user preference characteristics, the store characteristics and the commodity combination characteristics; The recommendation score corresponding to each initial product combination is determined according to the behavior information.

9. The method according to claim 1 or 8, characterized in that The selecting and recommending a target product combination from the multiple initial product combinations based on the recommendation scores includes: Sorting the multiple initial product combinations based on the recommendation scores and a preset sorting rule, and determining a preset number of target product combinations ranked first as the target product combinations; Push recommendation information corresponding to the target product combination to the terminal device of the target user.

10. A computer device, characterized in that: The computer device includes a controller and a memory, and the controller is used to implement the method for recommending a commodity combination as described in any one of claims 1 to 9 when executing a computer program stored in the memory.

Images