Independent Site Product Recommendation Method and Its Device, Equipment, Medium, Product

Abstract

The present application relates to an independent site product recommendation method and its device, equipment, medium, and product. The method includes: generating policy configuration information for an online store, where the policy configuration information includes the priority weights corresponding to each product recall policy in different business periods, and each product recall policy includes one or more recall channels; responding to a product recommendation request, obtaining the operation performance data of the current online store, and determining the business period according to the operation performance data; querying the corresponding policy configuration information according to the business period, and determining the effective recall channels under the target product recall policy corresponding to this business period; invoking the effective recall channels to perform a recall on the current online store, and sorting by matching the priority weights of the target product recall policy in the recall results to obtain a product recommendation list. The present application provides a unified solution for the policy configuration information of product recommendation for a cross-border e-commerce platform based on an independent site, and can improve the service experience.

Description

Technical Field

[0001] This application relates to the field of e-commerce information technology, and in particular, to a method for recommending products on an independent site, and a corresponding device, computer device, computer-readable storage medium, and computer program product. Background Art

[0002] Cross-border e-commerce is an industry that has developed rapidly in recent years. Independent sites are a new form of cross-border e-commerce. In operation, they have a high degree of autonomy, are more flexible, and avoid many platform rule restrictions. Compared with traditional e-commerce platforms, the types of online stores operating on independent sites vary, such as there are website group sellers and also boutique sellers; the scales of online stores are also different. Some stores have hundreds of products, and sometimes it may reach thousands or even tens of thousands. Due to the particularity of independent sites, different requirements are put forward for the product recommendation pages of independent sites compared with traditional e-commerce platforms. General product recommendation methods in the industry are no longer fully applicable in this scenario.

[0003] In a traditional product recommendation method, recommendations are made by combining the historical data of the customers to be recommended. Cluster the historical purchased products of the customers to obtain the interested categories, calculate the similarity between the products to be recommended and the categories, and recommend the products to be recommended with high similarity. This method is more suitable for platform-based e-commerce, where the customers visiting every day are old customers with historical behavior data, but it is not applicable to the e-commerce scenario of independent sites where the vast majority of customers are new customers.

[0004] Another product recommendation method based on the location information of customers will screen out N target users based on the location information and recommend the products purchased by the target users after processing. Although it solves the problem of new customers, the proximity in the dimension of geographical location does not mean that the preferences of these customers are the same. The recommendation strategy adopted by this recommendation method is too broad, and the obtained recommendation results are not accurate enough.

[0005] In addition to using behavioral data, there is also a product recommendation method based on product titles, which proposes to construct a knowledge graph for representation learning, calculate the title representation vectors of the query product and the candidate products, and recommend the candidate products with high matching probabilities. This method is a supplement to the lack of behavioral data, but the vector similarity of the titles cannot guarantee the accuracy of the matching.

[0006] It can be seen that in view of the particularity of independent sites, the above-mentioned recommendation methods are not individually applicable to bring effective product recommendation effects. Therefore, it is necessary to conceive solutions related to product recommendation separately from the perspective of the specific application field of e-commerce platforms based on independent sites. Summary of the Invention

[0007] The purpose of this application is to solve the above problems and provide an independent site product recommendation method and its corresponding device, computer device, computer-readable storage medium, computer program product. To achieve each purpose of this application, the following technical solutions are adopted:

[0008] On the one hand, to achieve one of the purposes of this application, an independent site product recommendation method is provided, including:

[0009] Generate the strategy configuration information of the online store. The strategy configuration information contains the priority weights corresponding to each product recall strategy in different business periods. Each product recall strategy includes one or more recall channels. The online store operates in an independent site of an e-commerce platform;

[0010] Respond to the product recommendation request associated with the current online store, obtain the operation performance data of the current online store, and determine the business period to which the current online store belongs according to the operation performance data;

[0011] Query the corresponding strategy configuration information according to the business period to which the current online store belongs, and determine the effective recall channels under the target product recall strategy corresponding to this business period;

[0012] Call the effective recall channels to recall the product information of the products on the shelves of the current online store, and sort the recall results by matching the priority weights of the target product recall strategy to which they belong to obtain a product recommendation list.

[0013] Optionally, generating the strategy configuration information of the online store includes:

[0014] Set the priority weights corresponding to each product recall strategy in different business periods according to the preset business periods, and represent the priority weights corresponding to each product recall strategy in each business period as a weight vector;

[0015] Construct a weight matrix according to the weight vectors corresponding to different business periods;

[0016] Respond to the strategy configuration request of the current online store, and configure the weight matrix as the default strategy configuration information of the current online store.

[0017] Optionally, setting the priority weights corresponding to each product recall strategy in different business periods according to the preset business periods includes:

[0018] Obtain the business stage information. In the business stage information, the business period is divided into a novice period, a rising period, and a mature period according to the operation duration of the online store from short to long;

[0019] During the novice period, set the priority weight of the first product recall strategy to the maximum value. The first product recall strategy is configured to include a recall channel for implementing semantic similar product recommendations based on the product information of the accessed product, and / or a recall channel for implementing similar product recommendations based on the product category of the accessed product;

[0020] During the growth period and the mature period, set the priority weights of multiple other product recall strategies to be greater than that of the first product recall strategy. The other product recall strategies are configured to include a recall channel for implementing product recommendations based on user behavior data, and the priority weight schemes of the corresponding product recall strategies in the growth period and the mature period are different.

[0021] Optionally, among the other product recall strategies, include:

[0022] The second product recall strategy, which is configured to include a recall channel for implementing co-purchased product recommendations based on the historical behavior types of the current user;

[0023] The third product recall strategy, which is configured to include a recall channel for implementing similar product recommendations based on the products corresponding to the real-time behavior of the current user;

[0024] The fourth product recall strategy, which is configured to include a recall channel for implementing popular product recommendations based on the click-through rate statistically obtained from the historical access behaviors of all users accessing the current online store.

[0025] Optionally, obtain the operation performance data of the current online store, and determine the business period to which the current online store belongs according to the operation performance data, including:

[0026] Obtain the operation performance data of the current online store, and the operation performance data includes the operation duration of the current online store and the daily user access volume;

[0027] Match the operation duration and the daily user access volume with the judgment conditions corresponding to each business period in the business stage information to determine the business period to which the current online store belongs.

[0028] Optionally, query the corresponding strategy configuration information according to the business period to which the current online store belongs, and determine the effective recall channels under the target product recall strategy corresponding to this business period, including:

[0029] Query the corresponding strategy configuration information according to the business period to which the current online store belongs, and determine the target product recall strategy and its corresponding priority weight therein;

[0030] Determine that the recall channels included in the target product recall strategy with a non-zero priority weight are effective recall channels for invocation.

[0031] Optionally, recall the product information of the products listed in the current online store through the effective recall channels, and sort the recall results by matching the priority weights of the target product recall strategies to which they belong, to obtain a product recommendation list, including:

[0032] Call the full-scale recall channels included in the target product recall strategy with a non-zero priority weight to obtain recall results, where each recall result includes multiple products and their recommended scores given by the recall channels;

[0033] Match the recommended scores of the products in each recall result with the priority weights of their corresponding target product recall strategies to obtain calculation results as sorting scores;

[0034] Sort the products in all recall results uniformly according to the sorting scores, and select some products with relatively high sorting scores to construct a product recommendation list.

[0035] On the other hand, to meet one of the objectives of the present application, there is provided an independent-site product recommendation device, including a policy configuration module, a period determination module, a channel determination module, and a recall processing module, where: the policy configuration module is used to generate policy configuration information for an online store, and the policy configuration information includes the priority weights of each product recall strategy corresponding to different business periods, and each product recall strategy includes one or more recall channels, and the online store operates in an independent site of an e-commerce platform; the period determination module is used to respond to a product recommendation request associated with the current online store, obtain the operation performance data of the current online store, and determine the business period to which the current online store belongs according to the operation performance data; the channel determination module is used to query the corresponding policy configuration information according to the business period to which the current online store belongs, and determine the effective recall channels under the target product recall strategy corresponding to this business period; the recall processing module is used to recall the product information of the products listed in the current online store through the effective recall channels, and sort the recall results by matching the priority weights of the target product recall strategies to which they belong, to obtain a product recommendation list.

[0036] On the other hand, to meet one of the objectives of the present application, there is provided a computer device, including a central processing unit and a memory, and the central processing unit is used to call and run a computer program stored in the memory to execute the steps of the independent-site product recommendation method described in the present application.

[0037] On the other hand, to meet another objective of the present application, there is provided a computer-readable storage medium, which stores a computer program implemented according to the independent-site product recommendation method in the form of computer-readable instructions, and when the computer program is called and run by a computer, it executes the steps included in the method.

[0038] In another aspect, a computer program product provided to meet another object of the present application includes a computer program / instructions. When the computer program / instructions are executed by a processor, the steps of the independent site product recommendation method described in any embodiment of the present application are implemented.

[0039] Compared with the prior art, the present application has multiple advantages, including but not limited to:

[0040] First of all, the present application uniformly provides policy configuration information for online stores in independent sites operating on an e-commerce platform. In the policy configuration information, multiple product recall policies are given according to the business periods to which the online stores belong. Each product recall policy may include one or more recall channels, so that the priority weights of the respective product recall policies corresponding to each business period can be independently configured. Thus, it is possible to flexibly provide effective product recall policies for online stores in different business periods, effectively avoiding the situation of inaccurate product recall caused by applying a single product recall policy. The e-commerce platform thus realizes the unified customization of a general product recall policy for all online stores under its platform, but each online store can flexibly apply specific product recall policies according to its own operation period and apply effective recall channels.

[0041] Secondly, a certain current online store can trigger a request through a configuration plug-in to implement a product recommendation service. After the current online store triggers a request, the e-commerce platform determines the operation period to which it belongs according to the operation performance data of the online store, and then determines the effective recall channels of the current online store according to the operation period, performs product recall according to the effective recall channels, and obtains a product recommendation list. Thus, each online store can call the product recommendation service of the e-commerce platform by configuring a standard plug-in or using a standard service. The development and maintenance costs of the e-commerce platform for realizing and opening the product recommendation service are relatively low, and the service experience can be improved.

[0042] In addition, the policy configuration information uniformly customized by the e-commerce platform includes the priority weights corresponding to each product recall policy. These priority weights can adjust the recommended scores given to the effective recall channels of the current online store, and thus realize the fine ranking of the product entries in the recall results, improve the accuracy of the product ranking after recall, obtain a more accurate product recommendation list, and improve the transaction matching efficiency between users and products, which can bring huge economic benefits to online stores and e-commerce platforms. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The above and / or additional aspects and advantages of the present application will become apparent and easy to understand from the following description of the embodiments in conjunction with the drawings, where:

[0044] Figure 1It is a schematic flowchart of a typical embodiment of the method for recommending independent-site products in this application.

[0045] Figure 2 It is a schematic logical structure diagram between the product recall strategy and the recall channel of the product recommendation service in the embodiment of this application.

[0046] Figure 3 It is a schematic flowchart of setting strategy configuration information for the current online store in the embodiment of this application.

[0047] Figure 4 It is a schematic flowchart of setting priority weights in different business periods in the embodiment of this application.

[0048] Figure 5 It is a schematic flowchart of sorting the recall results in the embodiment of this application.

[0049] Figure 6 It is a principle block diagram of the independent-site product recommendation device of this application;

[0050] Figure 7 It is a schematic structural diagram of a computer device adopted by this application. Detailed implementation manners

[0051] The embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals indicate the same or similar elements or elements with the same or similar functions from beginning to end. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be construed as a limitation to the present application.

[0052] Those skilled in the art of this technology can understand that, unless specifically stated otherwise, the singular forms "a", "an", "the" and "said" used herein may also include the plural forms. It should be further understood that the term "including" used in the specification of the present application means the presence of the described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or their groups. It should be understood that when we say that an element is "connected" or "coupled" to another element, it can be directly connected or coupled to other elements, or there may also be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The phrase "and / or" used herein includes all or any unit and all combinations of one or more related listed items.

[0053] Those skilled in the art of the present technology can understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as the general understanding of those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with the meaning in the context of the prior art, and will not be interpreted in an idealized or overly formal sense unless specifically defined as here.

[0054] Those skilled in the art of the present technology can understand that the "client", "terminal", and "terminal device" used herein include both devices with a wireless signal receiver that only has the ability to receive and no ability to transmit, and devices with receiving and transmitting hardware that can perform two-way communication on a two-way communication link. Such devices can include: cellular or other communication devices such as personal computers, tablet computers, etc., which have a single-line display or a multi-line display or a cellular or other communication device without a multi-line display; PCS (Personal Communications Service), which can combine voice, data processing, fax, and / or data communication capabilities; PDA (Personal Digital Assistant), which can include a radio frequency receiver, pager, Internet / intranet access, web browser, notepad, calendar, and / or GPS (Global Positioning System) receiver; conventional laptop and / or palm-held computers or other devices, which are conventional laptop and / or palm-held computers or other devices with and / or including a radio frequency receiver. The "client", "terminal", and "terminal device" used herein can be portable, transportable, installed in a vehicle (air, sea, and / or land), or suitable for and / or configured to operate locally, and / or operate in a distributed manner at any other location on the earth and / or in space. The "client", "terminal", and "terminal device" used herein can also be a communication terminal, an Internet access terminal, a music / video playback terminal, such as a PDA, MID (Mobile Internet Device), and / or a mobile phone with music / video playback function, or can also be a smart TV, a set-top box, etc.

[0055] The hardware referred to by the names such as "server", "client", and "service node" in this application is essentially an electronic device with the equivalent capabilities of a personal computer. It is a hardware device with the necessary components revealed by the von Neumann principle, including a central processing unit (including an arithmetic unit and a controller), a memory, an input device, and an output device. The computer program is stored in its memory, and the central processing unit loads the program stored in the external memory into the internal memory for execution, executes the instructions in the program, and interacts with the input / output devices to complete specific functions.

[0056] It should be noted that the concept of "server" in this application can similarly be extended to apply to the case of a server cluster. According to the network deployment principles understood by those skilled in the art, the various servers should be logically divided. Physically, these servers can either be independent of each other but can be invoked through interfaces, or integrated into a single physical computer or a set of computer clusters. Those skilled in the art should understand this flexibility and should not use it to restrict the implementation manner of the network deployment method of this application.

[0057] One or several technical features of this application, unless expressly specified, can either be deployed on the server and accessed by the client remotely invoking the online service interface provided by the server, or directly deployed and run on the client for access.

[0058] The neural network models cited or possibly cited in this application, unless expressly specified, can either be deployed on a remote server and remotely invoked on the client, or deployed on a client with sufficient device capabilities for direct invocation. In some embodiments, when it runs on the client, its corresponding intelligence can be obtained through transfer learning to reduce the requirements for the client's hardware operating resources and avoid excessive consumption of the client's hardware operating resources.

[0059] All kinds of data involved in this application, unless expressly specified, can either be remotely stored on the server or stored on the local terminal device, as long as it is suitable for being invoked by the technical solution of this application.

[0060] Those skilled in the art should be aware that although the various methods of this application are described based on the same concept and thus show commonality with each other, unless otherwise specified, these methods can all be executed independently. Similarly, for the various embodiments disclosed in this application, they are all proposed based on the same inventive concept. Therefore, for concepts with the same expression, as well as concepts that are only appropriately transformed for convenience although the concept expressions are different, they should be equivalently understood.

[0061] For each embodiment to be disclosed in this application, unless explicitly stated as mutually exclusive to each other, the relevant technical features involved in each embodiment can be cross - combined to flexibly construct new embodiments, as long as such combination does not deviate from the creative spirit of this application and can meet the requirements in the prior art or solve certain deficiencies in the prior art. Those skilled in the art should be aware of this flexibility.

[0062] An independent - site product recommendation method of this application can be programmed as a computer program product and deployed to run on a client or a server. For example, in an exemplary application scenario of this application, it can be deployed and implemented in the server of an e - commerce platform. Thus, by accessing the interface opened after the computer program product runs, human - machine interaction can be carried out with the process of the computer program product through a graphical user interface to execute this method.

[0063] Please refer to Figure 1 , in a typical embodiment of the independent - site product recommendation method of this application, it includes the following steps:

[0064] Step S1100: Generate the policy configuration information of the online store. The policy configuration information includes the priority weights corresponding to each product recall policy in different business periods. Each product recall policy includes one or more recall channels. The online store runs in an independent site of the e - commerce platform;

[0065] In the e - commerce platform in the application scenario of this application, it mainly manifests as a cross - border e - commerce platform, which allows each online store to run in a corresponding independent site and open services to the public users. The public users can access the page of the online store through various channels, including but not limited to the built - in page of the online store, the social media information display page, etc. The public users can directly reach the corresponding page of the online store through the links regarding these channels.

[0066] The server cluster of the e - commerce platform centrally maintains the background operation information of all online stores under the platform and provides various standardized services for these online stores, including but not limited to the product recommendation service of this application. When providing the product recommendation service of this application, each online store is allowed to enable the product recommendation service in the online store by configuring the plug - in corresponding to the product recommendation service provided by the e - commerce platform. Thus, when the public user accesses the online store and triggers a product recommendation request, a product recommendation list can be obtained for the public user through this product recommendation service to complete the product recommendation for the public user.

[0067] When an e-commerce platform obtains a product recommendation list in response to a product recommendation request, it can perform a product recall operation through one or more preset recall channels, so as to recall some of the listed products from the product information database of the corresponding online store according to the business logic of the recall channel, and construct the product recommendation list based on the product information of these listed products.

[0068] As Figure 2 As shown in the logical structure of , the recall channels can be divided and managed according to different product recall strategies, and thus can be set to include multiple product recall strategies. Each product recall strategy correspondingly includes one or more recall channels, and the recall channels included between different product recall strategies can be different. Similarly, the business logics for different recall channels to achieve product recall can also be different.

[0069] In different periods after an online store is put into operation, the number of its products, user traffic, and user behavior data will all show different characteristics. For example, in the early stage, the data generated during the operation process such as the number of products, user traffic, and user behavior data of the online store is relatively small, and it will increase correspondingly as the business time extends. The amount of data generated during the operation process usually affects the recall performance of each recall channel. For example, a recall channel that relies on user behavior data to recall products, in the early stage of the online store's operation, due to the lack of sufficient user behavior data in the online store, it will inevitably lead to the inability of this recall channel to perform a cold start to recall products with similar user behavior for a target product. Another example is that in the early stage of the online store's operation, similarly, it is also impossible to select the best among the listed products by counting the click-through rate of the listed products, thus affecting the effectiveness of the recall channel that uses this strategy for recall. To address these issues, multiple operating periods can be divided according to the characteristics of the data generated during the operation shown by the online store in different operating periods, and then the applicable logics of each product recall strategy can be determined correspondingly under different operating periods.

[0070] In one embodiment, the policy configuration information applicable to all online stores can be uniformly generated. In the policy configuration information, for each operating period, the applicable logic of each product recall strategy is respectively configured. In one embodiment, the applicable logic can be expressed as the priority weights corresponding to each product recall strategy. For example, the applicable logic corresponding to a certain operating period is set in the following form:

[0071] Operating period N: {Strategy 1: 2; Strategy 2: 3; Strategy 3: 1; Strategy 4: 0}

[0072] The expressions in the above examples indicate that the product recall policy 2 has the highest priority weight, followed by policy 1, then policy 3, and policy 4 has the lowest priority weight. The priority weights can be involved in subsequent operations. When the priority weight is 0, it can indicate that the policy is not enabled. For example, the above policy 4 is not enabled.

[0073] According to the principle of the above examples, after setting multiple operating periods in advance, the corresponding priority weights of each product recall policy can be set in the standardized policy configuration information under each operating period. Suppose there are 4 types of operating periods. Thus, an example of the organizational expression form of the standardized policy configuration information uniformly generated by the e-commerce platform is as follows:

[0074] Operating period 1: {Policy 1: 1; Policy 2: 0; Policy 3: 0; Policy 4: 0}

[0075] Operating period 2: {Policy 1: 1; Policy 2: 2; Policy 3: 3; Policy 4: 0}

[0076] Operating period 3: {Policy 1: 3; Policy 2: 2; Policy 3: 1; Policy 4: 1}

[0077] Operating period 4: {Policy 1: 1; Policy 2: 1; Policy 3: 1; Policy 4: 1}

[0078] It is not difficult to understand that by querying the policy configuration information according to the operating period in which each online store is located, the corresponding applicable logic during that operating period can be obtained, and the priority weights corresponding to each product recall policy can be obtained.

[0079] The e-commerce platform can set the policy configuration information as the default policy configuration information for the online stores operating on each independent site. For each online store that has enabled the product recommendation service, when it needs to recommend products to users, it can query the policy configuration information through the product recommendation service to determine multiple product recall policies and their corresponding priority weights corresponding to the operating period in which the online store is located, for generating a product recommendation list.

[0080] In one embodiment, the management users of each online store are allowed to edit the default policy configuration information by themselves to adjust the priority weights corresponding to each product recall policy, update the policy configuration information, and provide personalized custom recall services for the online store subsequently. The policy configuration information updated by an individual online store only serves the needs of that online store and does not affect other online stores from using the default policy configuration information uniformly generated by the e-commerce platform.

[0081] Step S1200, in response to a product recommendation request associated with the current online store, obtain the operation performance data of the current online store, and determine the operating period to which the current online store belongs according to the operation performance data;

[0082] When any public user accesses the product recommendation page of an online store, the online store is regarded as the current online store, and the product recommendation service plugin configured by the current online store will trigger a corresponding product recommendation request. This product recommendation request is sent to the background server of the e-commerce platform, and a corresponding product recommendation service process responds to the product recommendation request to determine a product recommendation list.

[0083] In response to the product recommendation request, the product recommendation service first obtains the operation performance data of the current online store. In one embodiment, the operation performance data may include the operation duration of the online store. The operation duration may be the difference in the number of days between the current time and the activation time marked in the store registration information database of the current online store on the e-commerce platform (usually the creation time of the online store). According to the length of this difference in the number of days, the business period to which the current online store belongs can be determined.

[0084] In one embodiment, business staging information for dividing different business periods is preset. For example, in the business staging information, four different business period types are set according to the judgment conditions less than or equal to 7 days, greater than 7 days and less than or equal to 30 days, greater than 30 days and less than or equal to 180 days, and greater than 180 days. Accordingly, by matching the difference in the number of days of the current online store with the judgment conditions of each business period type in the business staging information, the business period hit by the current online store can be determined according to the matching situation.

[0085] Step S1300: Query the corresponding policy configuration information according to the business period to which the current online store belongs, and determine the effective recall channels under the target product recall policy corresponding to this business period;

[0086] After determining the business period to which the current online store belongs, the corresponding policy configuration information of the current online store can be queried to determine the applicable logic of the product recall policy corresponding to this business time. For example, referring to the previous example, if the current online store is in business period 3, the example of the target product recall policy expression corresponding to it is as follows:

[0087] Business period 3: {Policy 1: 3; Policy 2: 2; Policy 3: 1; Policy 4: 1}

[0088] It can be seen from this that by determining the business period of the current online store, the corresponding product recall policies and the priority weights of each product recall policy in this business period can be obtained. Since each product recall policy includes one or more recall channels, in fact, the multiple recall channels applicable to the current online store when performing product recall in this business time are also determined. These recall channels constitute the effective recall channels of the current online store in the current business period.

[0089] In one embodiment, if there is a product recall policy with a priority weight set to 0 in the current business period, the recall channel corresponding to this product recall policy can be directly filtered out, that is, the recall channel under this product recall policy is not adopted to save the system overhead of the server. Of course, if not filtered, for the products recalled by this recall channel later, since the priority weight will be used to determine the sorting score of the product, the sorting score will be cleared to zero. Therefore, even if the recall channel performs a recall operation, the obtained recall results will not contribute when constructing the product recommendation list.

[0090] Step S1400: Invoke the effective recall channels to recall the product information of the products on the shelves of the current online store, and sort by matching the priority weights of the target product recall policies to which they belong in the recall results to obtain a product recommendation list.

[0091] After determining the effective recall channels of the current online store, the corresponding interfaces of each effective recall channel can be invoked to perform the product recall operation. Each effective recall channel executes the recall of the product information of the products on the shelves of the current online store according to its own inherent business logic to obtain corresponding recall results. The obtained recall results include the product information of each recalled product and the recommendation score given by the recall channel. Thus, the priority weight of the target product recall policy to which the recall result belongs can be multiplied by this recommendation score to obtain the sorting score of the corresponding product. Then, after merging all the recall results, all products are sorted according to the sorting score. Generally, a reverse sort can be performed so that the higher the sorting score, the more forward the product is. The product list obtained after sorting can be used as the product recommendation list corresponding to the product recommendation request and pushed to the graphical user interface of the public user who triggered the product recommendation request for display.

[0092] In one embodiment, a unified weight can be applied in the policy configuration information. For example, the weight of a product recall policy applicable to a certain business period is assigned a value of 1 to indicate activation, and the weights are uniformly represented as 1. The weight of a product recall policy not applicable to this business period is assigned a value of 0 to indicate non-activation. Thus, at the level of policy configuration information, each activated product recall policy no longer carries a differentiated weight, but only has a relative weight between the activated and non-activated product recall policies, and actually uses 1 and 0 as switch identifiers. For this situation, it is allowed to independently configure a weight matrix for the policy configuration information, and the weights corresponding to each product recall policy in each business period can be preset therein. This weight matrix can also be individually configured and maintained by the online store. Accordingly, the priority weight of each product recall policy in a business period is actually the product of the priority weight in the policy configuration information and the corresponding priority weight in the weight matrix. In this embodiment, the product recall policy and its weight are further decoupled, reflecting the flexibility of configuration and maintenance.

[0093] Based on the above embodiments, it can be seen that the present application has multiple advantages, including but not limited to:

[0094] First of all, the present application uniformly provides policy configuration information for online stores in independent sites operating on an e-commerce platform. In the policy configuration information, multiple product recall policies are given according to the business period to which the online store belongs. Each product recall policy can include one or more recall channels, so that each business period can independently configure the priority weights of its corresponding product recall policies. Thus, it is possible to flexibly provide effective product recall policies for online stores in different business periods, effectively avoiding the situation of inaccurate product recall caused by applying a single product recall policy. The e-commerce platform thus realizes unified customization of a general product recall policy for all online stores under its platform, but each online store can flexibly apply specific product recall policies according to its own operation period and apply effective recall channels.

[0095] Secondly, a certain current online store can trigger a request through a configuration plugin to implement a product recommendation service. After the current online store triggers the request, the e-commerce platform determines the operation period to which it belongs according to the operation performance data of the online store, then determines the effective recall channels of the current online store according to the operation period, and performs product recall according to the effective recall channels to obtain a product recommendation list. Thus, each online store can implement the invocation of the product recommendation service of the e-commerce platform by configuring a standard plugin or using a standard service. The development and maintenance costs of the e-commerce platform for implementing and opening the product recommendation service are relatively low, which can improve the service experience.

[0096] In addition, the strategy configuration information uniformly customized by the e-commerce platform includes the priority weights corresponding to each product recall strategy. These priority weights can adjust the recommendation scores given by the effective recall channels of the current online store, thereby realizing the refined ranking of the product entries in the recall results, improving the accuracy of the product ranking after recall, obtaining a more accurate product recommendation list, and enhancing the transaction matching efficiency between users and products, which can bring huge economic benefits to online stores and e-commerce platforms.

[0097] Based on any embodiment of the present application, please refer to Figure 3 , the step S1100 of generating the strategy configuration information of the online store includes:

[0098] Step S1110: Set the priority weights corresponding to each product recall strategy in different operating periods according to the preset operating periods, and represent the priority weights corresponding to each product recall strategy in each operating period as a weight vector;

[0099] In this embodiment, in the form of a vector, the effective representation of the strategy configuration information of the present application is realized. Therefore, for each operating period, the relationship information between each product recall strategy and its priority weight can be represented in the form of a weight vector. For example, the weight vector can be represented in the following form:

[0100]

[0101] According to the example here, it is not difficult to understand that each element in the weight vector corresponds to each product recall strategy in an orderly manner, and its element value represents the priority weight of the corresponding product recall strategy.

[0102] Step S1120: Construct a weight matrix according to the weight vectors corresponding to different operating periods;

[0103] For the weight vectors of different operating periods, further, the weight vectors corresponding to all operating periods can be represented by a data matrix, and its corresponding weight matrix can be determined. An example form is as follows:

[0104]

[0105] According to the example here, it is not difficult to understand that in the weight matrix, each row vector is the weight vector corresponding to an operating period, and each column position represents a specific product recall strategy, storing the priority weight corresponding to the product recall strategy.

[0106] Setting the strategy configuration information based on the matrix hardly occupies storage space and is convenient for efficient access. In the scenario where there are a large number of online stores on the e-commerce platform, such a storage strategy is particularly efficient.

[0107] Step S1130: In response to a policy configuration request of the current online store, configure the weight matrix as the default policy configuration information of the current online store.

[0108] For any current online store that configures the product recommendation service plugin, a corresponding policy configuration request will be automatically triggered when it configures the plugin. Thus, the background server of the e-commerce platform can configure the weight matrix as the default policy configuration information of the current online store.

[0109] In one embodiment, the background server of the e-commerce platform can set the policy configuration information for the online store without responding to the policy configuration request triggered by the online store. Instead, after generating the weight matrix, it is uniformly set as the standardized policy configuration information that all online stores can default to.

[0110] It is not difficult to understand from the above embodiments that representing the policy configuration information of this application in the form of a weight matrix basically does not occupy the system storage space, is convenient for efficient access, and can achieve standardized unified configuration and maintenance.

[0111] Based on any embodiment of this application, please refer to Figure 4 , the step S1110: Set the priority weights corresponding to each product recall policy in different business periods according to the pre-set business period, including:

[0112] Step S1111: Obtain business stage information, in which the business period is divided into a novice period, a growth period, and a maturity period according to the operation duration of the online store from short to long;

[0113] The e-commerce platform can preset the business stage information and set multiple business periods therein. For example, the business periods can include a novice period, a growth period, a maturity period, and other situations that do not belong to the above three periods. The novice period is used to represent the initial startup stage of the online store, the growth period is used to represent the adjustment and expansion stage of the online store, and the maturity period is used to represent the stable operation stage of the online store. It is not difficult to understand that the novice period, the growth period, and the maturity period usually correspond to different operation durations from short to long.

[0114] Step S1112: Adapt to the novice period and set the priority weight of the first product recall policy to the maximum value. The first product recall policy is configured to include a recall channel for implementing semantic similar product recommendation based on the product information of the accessed product, and / or a recall channel for implementing similar product recommendation based on the product category of the accessed product;

[0115] For an online store in its novice period, since there is not enough user behavior data generated in the store, some recall channels that rely on user behavior data are not suitable for adoption. Instead, a recall channel that implements semantic similarity product recommendation based on the product information of the accessed products is suitable. Such recall channels can include any one or all of two types. One type is to calculate the data distance between the semantic vectors of the product information of the target product and the semantic vectors of the product information of each product in the product information database of the online store, and select candidate products similar to the target product based on the data distance. The product information used for vector similarity matching usually includes any one or more of the product title, product attribute data, and product detail text. The other type can be to select products of the same category in the current online store's product information database based on the product information of the target product's category, and select the best from the selected products of the same category according to a certain condition, such as click-through rate. In this embodiment, the above-mentioned recall channels can be encapsulated as the first product recall strategy for centralized matching priority weights.

[0116] It is not difficult to understand that adopting the first product recall strategy for product recommendation for an online store in its novice period can solve the problem of insufficient user behavior data in the online store and the need for cold start recommendation of new products. Therefore, for an online store in its novice period, it is appropriate to set the priority weight of the first product recall strategy to the maximum value among all product recall strategies. Of course, the first product recall strategy can also be used in other business periods.

[0117] Step S1113: Adapt to the growth period and the maturity period, and set the priority weights of multiple other product recall strategies to be greater than the first product recall strategy. The other product recall strategies are configured to include recall channels that implement product recommendation based on user behavior data, and the priority weight schemes of the corresponding product recall strategies in the growth period and the maturity period are different.

[0118] For other business periods, including the growth period, the maturity period, and other situations in the previous examples, usually, an online store has generated a lot of user behavior data. Therefore, other product recall strategies can be set accordingly, and the priority weights of these product recall strategies can be set to be higher than the first product recall strategy. And, according to the actual situation, the priority weights of each specific product recall strategy can be set differently according to different business periods, such as the growth period and the maturity period.

[0119] For example, the other product recall strategies are configured to include recall channels that implement product recommendation based on user behavior data, and include:

[0120] The second product recall strategy is configured to include a recall channel for implementing co-purchased product recommendations based on the historical behavior types of the current user, mainly to achieve relevant product recommendations. For example, based on behavior types in user behavior data such as clicks, purchases, adding to the shopping cart, payments, etc., it determines whether the corresponding user has purchased a certain product. Accordingly, for each product with respect to each user, a behavior vector can be constructed to represent whether each user has accessed the product in a certain behavior type. Each product can obtain this behavior vector. For the target product, the data distance between its behavior vector and the behavior vectors of other products is calculated, and relatively similar partial products are selected according to the data distance, thereby achieving the recall of relevant products for the target product.

[0121] The third product recall strategy is configured to include a recall channel for implementing similar product recommendations for products corresponding to the real-time behavior of the current user, mainly to achieve personalized recommendations. For example, according to all the products accessed by the current user on the same day, products that are semantically similar to the product information of all the products currently accessed by the user are recalled from the product information database of the online store to achieve recall. Thus, relevant product recommendations can be provided according to the characteristics of the user's personal access behavior.

[0122] The fourth product recall strategy is configured to include a recall channel for implementing popular product recommendations based on the click-through rate statistically obtained from the historical access behaviors of all the accessing users of the current online store, mainly to achieve popular product recommendations. For example, an e-commerce platform can, based on the user behavior data generated by the current online store, statistically calculate the products with high access volume and order volume, sort them, and select some products with better access volume and order volume to achieve recall. Thus, fallback recommendations can be achieved.

[0123] It is not difficult to understand that for the rising period and the mature period, the priority weights of the second, third, and fourth product recall strategies can be different, and can be flexibly set by those skilled in the art according to experience.

[0124] According to the above embodiments, it can be seen that by combining the actual characteristics of the e-commerce field, the business period is set to include the novice period, the rising period, the mature period, etc., and the permissions of multiple product recall strategies in each business period are differentiated to adapt to each business period, so that each business period can comprehensively use multiple product recall strategies to achieve the best product recall effect, realizing the centralized and standardized management of a large number of recall channels on the e-commerce platform, and ensuring that the product recommendation service can effectively provide effective product recall services for online stores in different business periods.

[0125] Based on any embodiment of the present application, in step S1200, obtaining the operation performance data of the current online store and determining the business period to which the current online store belongs according to the operation performance data includes:

[0126] Step S1210: Acquire the operation performance data of the current online store, wherein the operation performance data includes the operation time of the current online store and the number of user visits on the day;

[0127] In this embodiment, each operating period and its corresponding judgment condition can be provided in the operating period information, so as to facilitate the determination of the operating period to which the online store belongs according to the judgment condition. The judgment condition of the operating period refers to two parameters for judgment. The first parameter is the operating time of the online store, and the second parameter is the number of user visits to the online store on that day. These two parameters are the operating performance data of the current online store. The operating time can be divided into multiple levels from short to long according to the artificially set time length for comparison and judgment; the number of user visits on that day can be compared with the average daily visits of the current online store within the recent historical time range to achieve judgment.

[0128] Step S1220: Match the operation time and the number of user visits on the day with the decision conditions corresponding to each operation period in the operation installment information to determine the operation period to which the current online store belongs.

[0129] The most recent time for counting the number of visits to Japanese yen can be the most recent week, i.e., 7 days. For an operating period, when the operating time of an online store and the number of user visits on that day both meet the judgment conditions corresponding to a certain operating period type, the corresponding operating period is confirmed to be the current operating period of the online store. In one embodiment, the operating period is set to include a novice period, a rising period, a mature period, and other situations. Therefore, the corresponding relationship between the operating period and its judgment conditions is exemplarily represented as follows:

[0130]

[0131] From the examples in the above table, it is not difficult to understand that the novice period is the initial stage of the online store, and its actual user traffic is limited; the rising period is the expansion period of the online store, and generally various adjustments are made, and its user visits are not stable enough. Therefore, in addition to the corresponding relationship between the operating time, the daily user visits of the online store during this period can be appropriately lower than the daily average visits; the mature period is the stable period after the online store has been in operation for a period of time. Therefore, as long as the operating time is greater than 30 days, and the daily user visits are roughly the same as the daily average, it can be confirmed that the online store has entered the mature period. For other situations that do not meet the judgment conditions of the first three operating periods, it indicates that there are abnormal phenomena and can be included in the "other situations" item to achieve a bottom-up treatment.

[0132] It is not difficult to understand from the above examples that the operation duration and the most recent period for statistically calculating the daily average number of visits are both empirical values or measured values that can be flexibly set. As long as the operation duration of the current online store, the number of user visits on the current day, and each judgment condition in the business staging information are matched, the corresponding business period of the current online store can be determined.

[0133] According to the above embodiments, it can be seen that as long as the operation performance data of the current online store is obtained and matched with the judgment conditions in the business staging information, the business period to which the current online store belongs can be quickly determined, which is convenient and efficient for determining the product recall strategy required for product recommendation.

[0134] Based on any embodiment of the present application, step S1300, query the corresponding policy configuration information according to the business period to which the current online store belongs, and determine the effective recall channels under the target product recall strategy corresponding to this business period, including:

[0135] Step S1310, query the corresponding policy configuration information according to the business period to which the current online store belongs, and determine the target product recall strategy and its corresponding priority weight therein;

[0136] After determining the business period to which the current online store belongs, by calling the policy configuration information for query, the corresponding weight vector can be determined, and the priority weights corresponding to each product recall strategy in this business period can be obtained. For example, in the previous example, is the weight vector corresponding to the mature period of the corresponding example, in which the corresponding priority weights for the first to the product recall strategies are set to 3, 2, 1, and 1 respectively.

[0137] Step S1320, determine that the recall channels included in the target product recall strategy with a non-zero priority weight are effective recall channels for invocation.

[0138] As described in the previous example, when the priority weight corresponding to the product recall strategy is 0, it means that this product recall strategy is not enabled. Therefore, in this embodiment, the product recall strategies with a priority weight can be filtered out first, and each recall channel in the product recall strategies with a non-zero priority weight can be determined as an effective recall channel, so as to only call the effective recall channels therein for product recall for the product recall strategies with a non-zero priority weight.

[0139] According to the above embodiments, it can be seen that in the process of implementing product recall, the effective recall channels available for the current online store can be quickly and efficiently determined through the policy configuration information, and the efficiency is very high.

[0140] Based on any embodiment of the present application, please refer to Figure 5, in step S1400, the effective recall channels are called to recall the product information of the products on the current online store, and the recall results are sorted by matching the priority weights of the target product recall strategies to which they belong, obtaining a product recommendation list, including:

[0141] Step S1410: Call the full-scale recall channels included in the target product recall strategies with non-zero priority weights to obtain recall results, and each recall result includes multiple products and their recommended scores given by the recall channels;

[0142] After determining the business period of the current online store and the corresponding priority weights of each product recall strategy in this business period, and removing the product recall strategies with zero priority weights, leaving the product recall strategies with non-zero priority weights and determining the recall channels among them as effective recall channels, the interfaces corresponding to these effective recall channels can be called to start the product recall operation, obtaining the recall results returned by each effective recall channel. It is not difficult to understand that each recall result includes multiple products and their recommended scores given by the recall channels.

[0143] Step S1420: Match the recommended scores of the products in each recall result with the priority weights of their corresponding target product recall strategies to obtain the calculation results as sorting scores;

[0144] Since the corresponding priority weights of each product recall strategy are given in the policy configuration information, this priority weight can be used to adjust the recommended scores in each recall result. Specifically, for the recommended scores of the products in each recall result, multiply them by the priority weight of the product recall strategy to which the recall result belongs to obtain the product as the sorting score.

[0145] Step S1430: Uniformly sort the products in all recall results according to the sorting scores, and select some products with relatively high sorting scores among them to construct a product recommendation list.

[0146] After determining the corresponding sorting scores for all the products in the recall results, all the recall results can be merged into a data set. In this data set, all the products are sorted according to the sorting scores of the products, usually in reverse order, and then a predetermined number of products with the highest sorting scores are selected. These products are constructed into a product recommendation list, and this product recommendation list can be used as the product recommendation result of the product recommendation service and pushed to the corresponding users of the current online store, completing the service process of recommending products for this user.

[0147] It can be understood from the above embodiments that the priority weights of each product recall policy in the policy configuration information of the present application can also play a role in sorting the product recommendation service, so that the product information recalled through multiple channels can be unified into the same dataset for sorting and optimization. Thus, a product recommendation list is screened and constructed to more accurately and comprehensively match the user's needs in the product recommendation list.

[0148] Please refer to Figure 6 , to provide an independent-site product recommendation device for one of the purposes of the present application, which is a functional embodiment of the independent-site product recommendation method of the present application. The device includes a policy configuration module 1100, a period determination module 1200, a channel determination module 1300, and a recall processing module 1400, where: The policy configuration module 1100 is used to generate the policy configuration information of the online store. The policy configuration information includes the priority weights of each product recall policy corresponding to different operating periods. Each product recall policy includes one or more recall channels. The online store operates in an independent site of an e-commerce platform; The period determination module 1200 is used to respond to a product recommendation request associated with the current online store, obtain the operation performance data of the current online store, and determine the operating period to which the current online store belongs according to the operation performance data; The channel determination module 1300 is used to query the corresponding policy configuration information according to the operating period to which the current online store belongs, and determine the effective recall channels under the target product recall policy corresponding to the operating period; The recall processing module 1400 is used to call the effective recall channels to recall the product information of the products on the shelves of the current online store, and sort the recall results by matching the priority weights of the target product recall policy to which they belong, so as to obtain a product recommendation list.

[0149] Based on any embodiment of the present application, the policy configuration module 1100 includes: a weight representation unit, which is used to set the priority weights of each product recall policy corresponding to different operating periods according to the preset operating periods, and represent the priority weights of each product recall policy corresponding to each operating period as a weight vector; a matrix construction unit, which is used to construct the weight vectors corresponding to different operating periods into a weight matrix; a response configuration unit, which is used to respond to the policy configuration request of the current online store and configure the weight matrix as the default policy configuration information of the current online store.

[0150] Based on any embodiment of the present application, the weight representation unit includes: an installment acquisition subunit, configured to acquire business installment information, in which the business period is divided into a novice period, a growth period, and a maturity period in ascending order of the operation duration of the online store; a novice configuration subunit, configured to adapt to the novice period and set the priority weight of the first product recall strategy to the maximum value. The first product recall strategy is configured to include a recall channel for implementing semantic similar product recommendation based on the product information of the accessed product, and / or a recall channel for implementing similar product recommendation based on the product category of the accessed product; other setting subunits, configured to adapt to the growth period and the maturity period, and set the priority weights of multiple other product recall strategies to be greater than the first product recall strategy. The other product recall strategies are configured to include a recall channel for implementing product recommendation based on user behavior data, and the priority weight schemes of the corresponding product recall strategies in the growth period and the maturity period are different from each other.

[0151] Based on any embodiment of the present application, among the other product recall strategies, there are included: a second product recall strategy, configured to include a recall channel for implementing co-purchase product recommendation based on the historical behavior types of the current user; a third product recall strategy, configured to include a recall channel for implementing similar product recommendation based on the product corresponding to the real-time behavior of the current user; a fourth product recall strategy, configured to include a recall channel for implementing popular product recommendation based on the click-through rate statistically obtained from the historical access behaviors of all accessed users of the current online store.

[0152] Based on any embodiment of the present application, the period determination module 1200 includes: a data acquisition unit, configured to acquire the operation performance data of the current online store, where the operation performance data includes the operation duration of the current online store and the daily user access volume; a period identification unit, configured to match the operation duration and the daily user access volume with the decision conditions corresponding to each business period in the business installment information, and determine the business period to which the current online store belongs.

[0153] Based on any embodiment of the present application, the channel determination module 1300 includes: a query determination unit, configured to query the corresponding policy configuration information according to the business period to which the current online store belongs, and determine the target product recall strategy and its corresponding priority weight therein; a channel filtering unit, configured to determine the recall channels included in the target product recall strategy with a non-zero priority weight as valid recall channels for invocation.

[0154] Based on any embodiment of the present application, the recall processing module 1400 includes: a recall execution unit configured to call the full-scale recall channels included in the target product recall policy with a non-zero priority weight to obtain a recall result, where each recall result includes multiple products and their recommended scores given by the recall channels; a sorting score unit configured to match the recommended scores of the products in each recall result with the priority weights of their corresponding target product recall policies to obtain a calculation result as the sorting score; and a list construction unit configured to uniformly sort the products in all recall results according to the sorting score, and select some products with relatively high sorting scores therefrom to construct a product recommendation list.

[0155] To solve the above technical problems, an embodiment of the present application further provides a computer device. As Figure 7 shown, it is a schematic internal structure diagram of the computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected through a system bus. Among them, the computer-readable storage medium of the computer device stores an operating system, a database, and computer-readable instructions. The control information sequence can be stored in the database. When the computer-readable instructions are executed by the processor, the processor can implement a product search category recognition method. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device can store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor can execute the independent site product recommendation method of the present application. The network interface of the computer device is used to communicate with the terminal. Those skilled in the art can understand that Figure 7 the structure shown in the figure is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0156] In this embodiment, the processor is used to execute Figure 6 the specific functions of each module and its sub-modules in the figure. The memory stores the program codes and various types of data required to execute the above modules or sub-modules. The network interface is used for data transmission between the user terminal and the server. The memory in this embodiment stores the program codes and data required to execute all modules / sub-modules in the independent site product recommendation device of the present application. The server can call the program codes and data of the server to execute the functions of all sub-modules.

[0157] The present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, one or more processors are caused to execute the steps of the independent site product recommendation method according to any embodiment of the present application.

[0158] The present application also provides a computer program product, including a computer program / instructions. When the computer program / instructions are executed by one or more processors, the steps of the method described in any embodiment of the present application are implemented.

[0159] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments of the present application can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the aforementioned storage medium can be a computer-readable storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0160] In summary, the present application provides a unified solution for the strategy configuration information of product recommendation for a cross-border e-commerce platform based on an independent site, enabling the e-commerce platform to standardize the configuration of the corresponding strategy configuration information for product recommendation for all online stores, enabling the online stores to query the strategy configuration information according to their business periods to determine their corresponding target product recall strategies, and implementing product recall according to the recall channels included in the target product recall strategies to generate a product recommendation list, completing an effective product recommendation process, improving the service experience, and ensuring the sound and efficient operation of the recommendation services of the online stores.

[0161] Those skilled in the art of the present technology can understand that the steps, measures, and solutions in the various operations, methods, and processes discussed in the present application can be alternated, changed, combined, or deleted. Further, the other steps, measures, and solutions in the various operations, methods, and processes discussed in the present application can also be alternated, changed, rearranged, decomposed, combined, or deleted. Further, the steps, measures, and solutions in the prior art that are the same as those disclosed in the various operations, methods, and processes in the present application can also be alternated, changed, rearranged, decomposed, combined, or deleted.

[0162] The above are only some embodiments of the present application. It should be noted that for those of ordinary skill in the art of the present technology, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. An independent site product recommendation method, characterized in that Including: Generating policy configuration information for an online store, where the policy configuration information contains the priority weights corresponding to each product recall policy in different business periods, and each product recall policy includes one or more recall channels, and the online store runs on an independent site of an e-commerce platform; Responding to a product recommendation request associated with the current online store, obtaining the operation performance data of the current online store, and determining the business period to which the current online store belongs according to the operation performance data; Querying the corresponding policy configuration information according to the business period to which the current online store belongs, and determining the effective recall channels under the target product recall policy corresponding to this business period; Invoking the effective recall channels to recall the product information of the products listed on the current online store, and sorting by matching the priority weights of the target product recall policy to which it belongs in the recall results to obtain a product recommendation list; The generating of the policy configuration information for the online store includes: Obtaining business stage information, and dividing the business period into a novice period, a growth period, and a maturity period in ascending order of the operation duration of the online store in the business stage information; Adapting to the novice period, setting the priority weight of the first product recall policy to the maximum value. The first product recall policy is configured to include a recall channel for recommending semantically similar products based on the product information of the accessed product, and / or a recall channel for recommending similar products based on the product category of the accessed product; Adapting to the growth period and the maturity period, setting the priority weights of multiple other product recall policies to be greater than the first product recall policy. The other product recall policies are configured to include recall channels for recommending products based on user behavior data, and the priority weight schemes of the corresponding product recall policies in the growth period and the maturity period are different.

2. The independent site product recommendation method according to claim 1, wherein Generating the policy configuration information for the online store includes: Setting the priority weights corresponding to each product recall policy in different business periods according to the pre-set business periods, and representing the priority weights corresponding to each product recall policy in each business period as a weight vector; Constructing a weight matrix according to the weight vectors corresponding to different business periods; Responding to a policy configuration request of the current online store, and configuring the weight matrix as the default policy configuration information of the current online store.

3. The independent site product recommendation method according to claim 1, characterized in that, Among the other product recall policies, including: The second product recall policy, which is configured to include a recall channel for recommending co-purchased products based on the historical behavior types of the current user; The third product recall policy, which is configured to include a recall channel for recommending similar products based on the products corresponding to the real-time behavior of the current user; The fourth product recall policy, which is configured to include a recall channel for recommending popular products based on the click-through rate statistically obtained from the historical visit behaviors of all accessed users of the current online store.

4. The independent site product recommendation method according to claim 1, wherein Obtaining the operation performance data of the current online store, and determining the business period to which the current online store belongs according to the operation performance data, including: Obtaining the operation performance data of the current online store, where the operation performance data includes the operation duration of the current online store and the number of user visits on the current day; Match the operation duration and the daily user access volume with the judgment conditions corresponding to each business period in the business stage information to determine the business period to which the current online store belongs.

5. The independent site product recommendation method according to claim 1, characterized in that Query the corresponding policy configuration information according to the business period to which the current online store belongs, and determine the effective recall channels under the target product recall policy corresponding to this business period, including: Query the corresponding policy configuration information according to the business period to which the current online store belongs, and determine the target product recall policy and its corresponding priority weight therein; Determine that the recall channels included in the target product recall policy with a non-zero priority weight are effective recall channels for invocation.

6. The independent site product recommendation method according to any one of claims 1 to 5, characterized in that, Invoke the effective recall channels to recall the product information of the products on the shelves of the current online store, and sort by matching the priority weight of the target product recall policy to which they belong in the recall results to obtain a product recommendation list, including: Invoke the full-scale recall channels included in the target product recall policy with a non-zero priority weight to obtain recall results, and each recall result includes multiple products and their recommended scores given by the recall channels; Match the recommended scores of the products in each recall result with the priority weights of their corresponding target product recall policies to obtain calculation results as sorting scores; Sort the products in all recall results uniformly according to the sorting scores, and select some products with relatively high sorting scores to construct a product recommendation list.

7. An independent site product recommendation device, characterized in that, Include: A policy configuration module for generating policy configuration information of an online store. The policy configuration information includes the priority weights corresponding to each product recall policy in different business periods. Each product recall policy includes one or more recall channels. The online store runs on an independent site of an e-commerce platform; A period determination module for responding to a product recommendation request associated with the current online store, obtaining the operation performance data of the current online store, and determining the business period to which the current online store belongs according to the operation performance data; A channel determination module for querying the corresponding policy configuration information according to the business period to which the current online store belongs, and determining the effective recall channels under the target product recall policy corresponding to this business period; A recall processing module for invoking the effective recall channels to recall the product information of the products on the shelves of the current online store, and sorting by matching the priority weight of the target product recall policy to which they belong in the recall results to obtain a product recommendation list; Among them, the policy configuration module includes: A stage acquisition subunit for acquiring business stage information, in which the business period is divided into a novice period, a rising period, and a mature period according to the operation duration of the online store from short to long; A novice configuration subunit for adapting to the novice period, setting the priority weight of the first product recall policy to the maximum value. The first product recall policy is configured to include a recall channel for implementing semantic similar product recommendation based on the product information of the accessed product, and / or include a recall channel for implementing similar product recommendation based on the product category of the accessed product; An other setting subunit, which is used to adapt to the rising period and the mature period, and sets the priority weights of multiple other product recall policies to be greater than that of the first product recall policy. The other product recall policies are configured to include recall channels for implementing product recommendations based on user behavior data, and the priority weight schemes of the corresponding product recall policies in the rising period and the mature period are different.

8. The independent site product recommendation device according to claim 7, wherein Among the other product recall policies, there are included: A second product recall policy, which is configured to include a recall channel for implementing co-purchase product recommendations based on the historical behavior types of the current user; A third product recall policy, which is configured to include a recall channel for implementing similar product recommendations based on the products corresponding to the real-time behavior of the current user; A fourth product recall policy, which is configured to include a recall channel for implementing popular product recommendations based on the click-through rate statistically obtained from the historical access behaviors of all visiting users of the current online store.

9. A computer device, comprising a central processing unit and a memory, characterized in that, The central processing unit is used to call and run the computer program stored in the memory to execute the steps of the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, It stores in the form of computer-readable instructions a computer program implemented according to the method according to any one of claims 1 to 6. When the computer program is called and run by a computer, it executes the steps included in the corresponding method.

Images